Perceived safety in public spaces

A quantitative investigation of the spatial and social

influences on safety perception among young adults in
Stockholm

Walt Machielse

June 2015
Supervisor: Ulf Jansson
Department of Human Geography
Stockholm University
SE-106 91 Stockholm / Sweden
www.humangeo.su.se
Abstract

Machielse, Walt (2015). Perceived safety in public spaces: a quantitative investigation of the spatial
and social influences on safety perception among young adults in Stockholm

Human Geography, master thesis for master exam in Human Geography, 30 ECTS credits

Supervisor: Ulf Jansson

Language: English

Key Words: Quantitative, Safety, Young Adults, Spatial, Social, Influences, Perceived, Seasons.

From existing literature a distinction between young and older people has been established
regarding safety perception in public spaces. This quantitative study focuses on the young adult
group in order to further investigate the reasoning and behavioral choices young adults make in
relation to using public spaces and safety perception. In order to do so, the main research question
is: in what ways does safety perception affect the use of public spaces by young adults? Using the
quantitative method of surveying and statistical analysis, several hypotheses were developed and
answered. The main outcome from the statistical research is that within the young adult group, there
are several groups that have significantly different opinions and behavior regarding safety
perception. Seasonal influence, mostly based upon the lack of daylight, is stronger among women
than men. Women feel significantly safer during the summer than men. This increased feeling of
safety among women during the summer has a positive influence on the use of public spaces by this
group. Seasonal influences and darkness have significantly less effect on frequent visitors to the city
of Stockholm and people that have lived for a relatively long time in Stockholm. This confirms the
statements of the existing literature regarding men and women in relation to safety perception.

2
Contents

Abstract P. 2

1. Introduction P. 5
 1.1 Aim P. 6

2. Theoretical framework P. 7
 2.1 Introduction P. 7
 2.2 Public spaces P. 7
 2.3 Safety and darkness P. 8
 2.4 Security and maintenance P. 9
 2.5 Environmental effects on safety P. 10
 2.6 What does it mean to feel unsafe? P. 11
 2.7 Conceptual model P. 11
- 2.7.1 Personal characteristics P. 12
- 2.7.2 Perception public space P. 13
- 2.7.3 Degree of perceived safety P. 14

3. Methodology P. 16
 3.1 Introduction P. 16
 3.2 Sub-questions and hypotheses P. 16
 3.3 Quantitative research: survey and statistics P. 17
 3.4 Target group P. 18
 3.5 Research area P. 18
 3.6 Heat maps P. 20
 3.7 The survey in detail P. 21
- 3.7.1 General sense of safety – Part 1/3 P. 21
- 3.7.2 Feeling unsafe in the city – Part 2/3 P. 23
- 3.7.3 Personal characteristics – Part 3/3 P. 23
 3.8 Ethical issues P. 24
 3.9 The process of surveying P. 25

3
4. Representativeness and descriptive statistics P. 27
 4.1 Calculating the sample size and the minimum of responses needed P. 27
 4.2 Descriptive statistics: characteristics of the study P. 29
- 4.2.1 Gender P. 30
- 4.2.2 Age P. 30
- 4.2.3 Ethnicity P. 31
- 4.2.4 Amount of time lived in or near Stockholm P. 32
- 4.2.5 Employment level of the respondents P. 32
- 4.2.6 Educational level P. 33
- 4.2.7 Public spaces and feeling unsafe P. 34
 4.3 Representativeness P. 34
- 4.3.1 Gender P. 35
- 4.3.2 Age P. 35
- 4.3.3 Ethnicity P. 35

5. In-depth analyses of the hypotheses and heat maps P. 37

 5.1 Hypothesis 1 P. 37
 5.2 Hypothesis 2 P. 38
 5.3 Hypothesis 3 P. 40
 5.4 Hypothesis 4 P. 41
 5.5 Heat map results P. 43
 5.6 Extra testing P. 46
- 5.6.1 Safety perception in different seasons P. 46
- 5.6.2 Visiting frequency to the city P. 47

6. Conclusion P. 48

7. Discussion P. 51

References P. 54

8. Appendix P. 56
 8.1 Research area P. 56
 8.2 Survey P. 57
 8.3 Representativeness tests P. 67
 8.4 In-depth tests P. 69

4
1. Introduction

By offering publically accessible space which connects other areas of the city, public spaces are vital
for the existence of a well-functioning city. However, the quality and attractiveness of public spaces
can differ, as well as the intention of the space by urban planners. The way people perceive space
determines the way space is used. One of the major factors that influences people’s behavior is
sense of safety. If people feel unsafe in an area, the behavior of people and sense of comfort is
changed. The way people perceive their safety can tell a lot about the way people make use of
certain parts of the city. This information can be used for future urban planning projects regarding
public spaces, in order to make places feel more safe, comfortable and popular. Raising awareness
about factors that people find most important regarding their safety in public spaces in combination
with looking at reported crime data can determine if there is a correlation between actual crimes and
the perception people have about crime. Collecting data about perceived safety levels regarding day
and night time can also show which environmental factors are most important in building a general
sense of safety.
Public space is a concept that can be interpreted in different ways. For this thesis, public
spaces will be treated as spaces that are publicly accessible (either privately or publicly owned). The
sense of safety will be derived from the literature and compiled into a conceptual model. Building a
theoretic model containing the elements of safety that people value when using public spaces will
offer the platform for the quantitative study. The target group of this study will be young adults,
mainly because young adults tend to be more active in the nightlife scene, and therefore making use
of public space both during day and night time, which could give an insight in the influence of
daylight on safety perception.
A survey from the Swedish National Council for Crime Prevention (Brå, 2014), states that men
are less likely to worry about crime in general than women. Additionally, the younger age groups
tend to worry less than people in the older age groups. Using young adults for this study may offer
more insight into the reasons why this group worries less than older people, and the divide between
young women and men could be looked at into detail.
The selected areas for research are determined by looking at their environmental properties;
spatial features that are most relevant regarding safety of visitors of public spaces will be taken into
account when selecting research areas for this study. The existing literature regarding safety in public
spaces mentions for example the presence of streetlights (Shenassa et al, 2006). For a city as
Stockholm where darkness plays a big role during the winter months, the importance of streetlights
will most likely be one of the factors people value a lot in this study. What this means is that the
origin of studies must be acknowledged and taken into account when conducting a similar study;
environmental factors are of great importance, some more than others depending on the research
area.

To determine the value and effects safety has on the use of public spaces the following research
question has been made:
In what ways does safety perception affect the use of public spaces by young adults?

Additionally, several sub questions will help formulate hypotheses, in order to make statistical
research possible.

5
1.1 Aim
The aim of this thesis is to look into the perception of safety of public spaces among young adults in
Stockholm. Using quantitative methods to gain empirical results, statistical analysis can offer insight
regarding the factors that affect the perception of safety. The selected target group for this study has
the qualities to have an opinion about the selected public spaces both during the day and night, due
to the assumption that young adults would make use of public spaces in the city both during day and
night time. The results of this study could offer potential solutions to the planning of (urban) public
spaces that would increase the perceived safety, by applying the factors that become apparent in this
study. People’s sense of safety can affect day to day activities and change the way people live their
lives. The outcomes of this thesis could show insight into environmental factors that affect public
spaces negatively and which consequently, makes people avoid them. If these environmental factors
can be changed, certain public spaces of the city can be made more attractive and will increase the
general perceived safety among people. Additionally, registered crimes in public spaces and their
locations will be compared to the places that are perceived as dangerous, from the empirical study,
to see if there is a correlation between factual and subjective information.

6
2. Theoretical framework

2.1 Introduction
This chapter discusses the information from previous studies on the topic of safety perception and
public spaces. The goal of this chapter is to clarify concepts, theories and questions that might arise
from the literature. Additionally, this chapter is the ground work for the quantitative research and
eventually the results chapter of this thesis. Having a theoretical framework is essential and needed
for conducting either quantitative or qualitative research. The information from the previous studies
will be transformed into a conceptual model, which will contain the elements that revolve around
safety perception in public spaces. The conceptual model for this thesis will be discussed at the end
of this chapter, after discussing the factors that influence safety perception.

2.2 Public spaces

Public spaces are built to offer places to socialize, work and travel through. It is a broad concept that
suggests ‘openness’ of space; open for anyone who wishes to make use of it. There are quite a lot of
definitions of public space, such as “space that is not controlled by private individuals or
organizations, and hence is open to the general public” (Mehta, 2014, p. 54). For this thesis,
ownership of the space is not that relevant. What’s more relevant is the use and accessibility of the
space by the public, for example, a park can be owned privately but still be open to the public, in this
thesis this would still be classified as a public space. When discussing public spaces in a city, one
could claim that almost everything in the city is public space, including the roads crossing through the
city. This thesis limits itself to squares, plazas, parks, fields, harbor areas and sidewalks. This means
that privatized areas that are nowadays considered to be public spaces, for example malls, are not
considered part of public space within this thesis.
The history of the public space is related to democracy, stemming from the old Greek agora.
The first public spaces were meant to function as places where society came together to make
decisions. It was a space created for the people in order to voice their opinion on governmental
issues. Additionally it was also a market space, which led to the specialization of products and trade
between these products. The market space, a public space, was a medium for the trade of specialized
goods. The original idea of public spaces was to offer a space to bring people together for business
and social activities. Nowadays public spaces such as parks and squares are still used for such
purposes. Flea markets, food markets and clothing markets are the example of businesses conducted
in public spaces. Public spaces usually also offer some sort of platform for social activities such as
playgrounds, benches and tables for picnics, sport areas (such as goal posts on grass fields) and
walking paths (Mehta, 2014). According to Thomas (1991) public spaces have four major functions
nowadays: they function as a space for public life to flourish (1), as a place for different social groups
to interact or coexist (2), to facilitate a space for the display of symbols and images in society (3) and
to tie other urban spaces together (4). The last function mentioned by Thomas (1991) is the only one
that is not related to social matters, but instead to urban design. However, the three other functions
show how important public spaces are for society to express itself. Public spaces offer a social
platform shared by different social groups for people to connect and share information which is an
important feature of expressing democracy (Mehta, 2014; Arendt, 1958).
There are several categories to determine the quality and attractiveness of public spaces, one
of them being safety. Safety is considered the most important factor when valuing public spaces,
because the perceived safety of an area by the individual has a strong influence in the decision by the
individual to make use of the space, or to avoid it (Mehta, 2014, p. 60). The perceived safety within
public spaces of an individual is dependent on several factors, such as: perceived security,
maintenance of the area, visibility in the area, the presence of green (bushes, trees, grass, flowers
etc.), the presence of water, streetlights, the number of people visiting the area, and time of day

7
(Schroeder & Anderson, 1984, pp. 181-187; Mehta, 2014, p. 60; Shenassa et al., 2006, pp. 1015-
1016).
A public space that is perceived as safe could potentially be experienced as comfortable. The
individual perceives the area as non-threatening, which positively influences the level of comfort the
person experiences (Mehta, 2014, pp. 60-61). However the comfort level of a public space is not
necessarily defined by the level of safety. Different seasons, for example, define the use of the space
and for a large part influence the level of comfort. Sunlight is an important actor in relation to the
use of public spaces. During warm months, people tend to make more use of public spaces than in
colder months, because the temperature and environment allows people to do so. Recreational
facilities such as benches, playgrounds and water fountains become attractive and usable when it is
warm and sunny outside, adding to the esthetic value of the public space (Mehta, 2014). Even
though public space is open for everyone to make use of, certain groups that are making use of the
space can make people, such as youths, homeless people or drug addicts hanging around in that area
which would make it less attractive for elderly for leisure activities. So even though public space in its
definition claims to be accessible for everyone, areas tend to divide themselves into ‘go’ and ‘no-go’
areas.

2.3 Safety and darkness

Within the study done by Shenassa et al. (2006), the perceived safety of the neighborhood of the
respondents was determined by asking if the respondents felt safe returning to their dwellings in the
dark. The factor of night time could be further expanded within this study by asking respondents if
there are differences between safety perception during different seasons (for example: one could
feel safe in a park during the summer because of the longer availability of daylight, but this does not
mean the individual would feel safe in the same park during winter time when daylight is more
limited). In a study from Bennett et al. (2007) about neighborhood safety and physical activity, the
difference between men and women was discussed; the level of own safety was determined by
asking the participants if they felt comfortable walking alone in their neighborhood during day- and
night time. Bennett et al. (2007) concludes that safety perception depends on the time of day. The
time of day plays an indispensable role when studying the safety perception among individuals.
Women appear to have an increased sense of feeling unsafe during night time compared to men
(Bennett et al., 2007; Shenassa et al., 2006).
In a night time setting people are more likely to be attentive to other people and their
surroundings, shaped by the fear the something might happen to them. Public spaces that are well lit
are less likely to be perceived as unsafe, mostly because light gives people the ability to get an idea
and overview of a space. If a space is not well lit and visibility is limited, an increased sense of feeling
unsafe is triggered (Brands et al., 2015, p. 442). However, a well-lit area is also associated with less
potential crime that could occur; people feel that sufficient lighting will force crime to move
elsewhere and that well-lit areas therefore create a sense of safety (Brands et al., 2015, p. 449).
In a literature study from Maruthaveeran & Konijnendijk van den Bosch (2014) lighting was
mentioned as an important factor that affects the perceived safety. However, the study made clear
that the presence of lighting in a public space does not necessarily guarantee a feeling of safety
among visitors. Physical attributes of spaces combined form the perceived safety; lighting and lots of
green in an area (such as high bushes) combined made people feel less uncomfortable. People seem
to be more comfortable with low cut green areas, which gives visitors more visibility of the entire
area (Lindgren & Nilsen, 2011, pp. 202-203). There are also exceptions to the general rule that some
form of lighting works positively in an individual’s safety perception: one case showed that walking
through a park with dark paths felt safer than when these paths were accommodated with
streetlights, because the darkness made the individual feel less visible, and thus less likely to become
a victim of a crime (Maruthaveeran & Konijnendijk van den Bosch, 2014, p. 4). Related to darkness,

8
the time of day stands in relation to esthetic and safety perceptions; an area that during day time
(when there is plenty of daylight) is perceived as disorderly will be perceived even more disorderly
during night time (Liempt, van et al., 2015, pp. 407-408). The difference between day and night time,
regarding types of crimes, is that fear of night time crimes are mostly sexual assault crimes while fear
of day time crimes is much more diverse (Maruthaveeran & Konijnendijk van den Bosch, 2014, p. 13).
It must be noted here that there is a difference between men and women and the fears about
different types of crimes; for example, women are more likely to worry about sexual assault crimes
than men (Brå, 2014; Maruthaveeran & Konijnendijk van den Bosch, 2014).

2.4 Security and maintenance

Feeling secure in a public area can be achieved by having surveillance cameras or police patrolling
the area, but by overdoing this, the idea that the area instead is unsafe might arise, and the police
and visible security in the area are needed because otherwise crime would prevail (Mehta, 2014,
p.60). There is a connection between visitors of the public space and security; if there are enough
people within the public space, a form of self-securitization is created where people keep an eye on
each other and on potential cases of law-breakers. This would mean that the visible ‘organized’
security is not needed and perhaps even unwanted. Maintenance of public spaces regards the
cleanliness of an area, including proactively preventing vandalism or repairing damaged property in
the public space. Vandalism, such as graffiti or broken attributes in the area, is an important factor
that determines the esthetic value of an area; vandalized areas are less attractive and keeps people
away (Shenassa et al., 2006, p. 1015; Perkins et al., 1992, p. 29). The broken window theory can be
applied here: if small crimes such as vandalism are prevented, people perceive the area as neat and
orderly, and worse crimes will have less of a chance happening (Gau et al., 2014, p. 579).

There is a relation between vandalism and a ‘fear of disorder’ among people. Perceived vandalism or
unlawfulness gives people the idea that there is no control in the area and that people visiting the
public space are vulnerable to (witnessing) crime themselves (Gau et al., 2014, p.580). This fear of
crime is a concept that is mostly initiated and spread by media reports and through stories that
people hear from other people. Even though these stories might for a part be untrue, they still have a
strong effect on the way people perceive public spaces (Brands et al., 2015, p. 440). Feeling unsafe
can manifest itself in many ways, for example starting to feel physically ill, uneasy with increased
awareness of being in an unsafe environment, or the preventative instinct by estimating the
potential danger and avoiding the unsafe area. There are many emotions that come into play
regarding fear or crime. People learn from other people or from watching the news that bad things
happen, in some parts of a city more than in other places. People make mental notes of these
learning moments and remember what they’ve seen or heard when they are about to pass by similar
areas (Brands et al., 2015, pp. 440-443).
Expanding further on the broken window theory, the lack of a well maintained public space
is linked to increase the uneasiness people experience in an area. Additionally, a public space that
has visible security in the form of surveillance camera’s or police patrolling is found to be safer and
therefore more attractive. However the presence of security in the form of policemen has two sides
to it (Mehta, 2014). On the one hand people find policemen in public space comforting and
contributing to the perception of safety of an area, on the other hand people relate policemen to
trouble or potential danger. People start to wonder if the public space secured by policemen is a
potential danger area, and it starts to have the counter effect of what was actually desired in the first
place with visible security in the area (Brands et al., 2015, pp. 450-451; Gau et al., 2014, pp. 580-581).
The alternative to official security within public spaces is securing space by community. This form of
security is employed by the people that make use of the public space for recreational, social or
business purposes; it is the basic concept of keeping an eye on each other, and therefore leaving

9
little room for crime to happen. The environment in which crime could potentially manifest itself also
depends on the location and type of public space. A park for example, characterizes itself by lots of
green in combination with ill lit areas. Spaces such as these are difficult to monitor with surveillance
cameras and are probably better secured by the visitors of the space itself (Schroeder & Anderson,
1984; Mehta, 2014, pp. 60-62).

2.5 Environmental effects on safety

Environmental factors, which are decided when planning and building public spaces, such as
streetlights, green areas and general visibility in the area form the atmosphere of an area. For
example, a park with lots of small paths and lots of green during the day can create a romantic
atmosphere, but this same park could be perceived as dangerous during the night if it lacks the
needed streetlights for people to see what is going on around them. The amount of trees, bushes
and maintenance of green areas can influence the perceived safety of people and the preferences for
visiting or stay clear of a public area (Kuo et al., 1998, p. 33-35). These environmental factors are
created during the planning stage of the area, and therefore have the first and direct influence in
creating a sense of safety.
The design of a public space is done with a specific purpose in mind. On the one hand public
spaces can be quite inviting, with benches alongside a park trail for example. On the other hand,
often times benches are designed in such a way that people cannot lie down on them, to prevent
unwanted people or groups from making use of them (The Guardian, 2014). These new urban design
methods are applied in spaces that have had reports of homeless people, and try to make it
impossible for the homeless to survive in the newly designed spaces. When looking at the outcome
of these new designs fewer homeless people will be present in the new areas. One could question
the outcomes of these ‘defensive urban design’ methods, because it does not address the actual
problem of homelessness, it just moves the problem to other spaces. However, some aspects of
defensive urban design do contribute to clearing certain public spaces from beggars and homeless
people, which adds to potential increased safety levels among individuals visiting the public spaces. A
study conducted by Ferraro & LaGrange (1992) discusses the biggest fears regarding crimes that
could potentially happen to the respondents. The interaction with a beggar is measured within the
study, concluding that women are more likely to be afraid of being approached by a beggar on the
streets (Ferraro & LaGrange, 1992, p. 241). The broken window theory mentioned before has a major
influence on the way people perceive their environment and therefore, their own safety. Areas that
contain graffiti drawings are associated with a lack of maintenance and surveillance. Vandalism in the
form of graffiti can give people the idea that these kinds of illegal activities are not prevented or
addressed, which could lead to questioning their own safety. There are many cases in which the
spreading of ‘green’ in public spaces can lead to a decrease in vandalism and crime. For example, in a
certain area of Philadelphia, the cleaning up and planting new trees and grass fields in previously
empty spaces helped reduce crime by 90%. This example shows that the methods used within the
broken window theory can be applied to different situations and achieve similar goals, by applying
‘natural security’. Similar to providing plenty of trash bins which prevents people from throwing trash
on the ground, the planting of trees, grass and flowers prevents people from vandalizing or spraying
graffiti on walls (Kuo et al., 1998, p. 31). A study conducted by a Dutch research firm concludes that
young people actually prefer green spaces to non-green spaces. According to the research firm
stichting wAarde, young people like green spaces because it allows them to ‘find peace and to clear
their heads’, and that green areas reduce their stress levels (NRC, 2015). The firm does not report on
perceived safety within green spaces such as parks or forests, and implementing new green in urban
areas would not be a good idea if its visitors would not feel safe. The presence of green and the
presence of too much green is a difficult discussion, but nevertheless important when looking at the

10
perceived safety among visitors (Maruthaveeran & Konijnendijk van den Bosch, 2014; Lindgren &
Nilsen, 2011).

2.6 What does it mean to feel unsafe?

Feeling unsafe and being unsafe are two different concepts, which can differ from each other. Being
unsafe refers to facts regarding crime; if there are a lot of registered offenses in a particular area, the
area can be classified as unsafe. It is important to stress the objectivity of this concept, there are no
individual perceptions or feelings that have anything to do with ‘being unsafe’ (Ruijsbroek et al.,
2015, pp. 39-40). Feeling unsafe, however, is an emotional concept. For example, an individual can
get the idea that an area might be unsafe, while according to the facts the area is perfectly safe.
Feeling unsafe is a concept that can consist of many factors and emotions. The fear of
becoming a victim of a crime (for example being mugged) is different than the fear (or anxiety) of
witnessing public harassment, or seeing vandalized parts of an area (Ruijsbroek et al., 2015, p. 40;
Pain, 2000). It must be noted here that even though fear can be classified into many categories, the
way fear is perceived and experienced by the individual can differ from person to person, which is
exactly why this is a sensitive topic.

Experiencing fear in public spaces has several negative effects on the individual, one of which is
especially relevant for this thesis: ‘fear of the other’. The fear of crime and anxiety towards other
negative experiences has a negative impact on the levels of trust that an individual has towards other
people (Ruijsbroek et al., 2015, p. 40). The impacts that perceived fear can have on people and the
declined trust of ‘the other’ could result in a decrease in social cohesion. As stated in the beginning
of this chapter, the main goal of public spaces was to offer a space for people to get together and
socialize. However, the perceived fear of individuals in public spaces could put social cohesion at risk
(Stafford et al., 2007). The relation between social cohesion and perceived fear levels correlates
negatively, which means that if social cohesion levels are high in a certain area, people are less likely
to feel unsafe. The levels of social cohesion in public spaces are dependent on the time of day and
seasons of the year; people are more likely to socialize in public spaces when it is light outside and
warm (during the spring/summer months) (Mehta, 2014, p. 61). Maruthaveeran & Konijnendijk van
den Bosch (2014) classify fear or anxiety in public spaces under two categories: social related fears
and physical related fears. The first category discusses the before mentioned ‘fear of the other’;
seeing strangers litter or vandalize a part of an area have strong effects on one’s safety and the fear
of the other, these are examples of social related fears. Physical related fears are caused by the
environment, such as darkness (lack of lighting), the presence of green in an area, and the overall
esthetic value of an area. Maruthaveeran & Konijnendijk van den Bosch (2014) state that social
related fears are more important than physical related fears, however one could argue that these
categories cannot be judged separately. The fear that someone might hurt an individual might be a
social related factor, but this fear could become especially relevant when the individual travels
through an area that is perceived as unsafe by the individual, which would be a physical related
factor (Ferrero & LaGrange, 1992; Bromley & Stacey, 2012).
The consequences of feeling unsafe can have negative effects on an individual’s personal life,
but also on the health of the individual. Someone that has to deal with a lot of areas that make an
individual feel unsafe has elevated stress levels, which deteriorates the overall health (Stafford et al.,
2007).

2.7 Conceptual model

The before-mentioned theory on which factors influence safety perception such as: darkness, the
esthetic value and maintenance of an area, the amount of green in an area, each play vital roles in
shaping one’s safety perception. However, it is necessary to note that these factors shape a network

11
and influence each other in the process that is safety perception of the individual. For example
‘darkness’ alone does not mean much if it is not put into some sort of context. A dark park is
something people are able to relate to, and are therefore able to create an idea of safety about.
A conceptual model takes the most relevant conclusions from the theoretical chapter and illustrates
the relationship between different actors and/or processes. The use of a conceptual model in this
thesis is to clarify the factors that could potentially have an influence on the degree of perceived
safety by young adults.

2.7.1 Personal characteristics

The individual’s characteristics consisting of age, gender, daily occupation and the amount of times
they travel through the selected research area within a given time will, for a large part, shape an
individual’s view on their own safety in public spaces (Bennett et al., 2007; Brands et al., 2015). Using
arrows in the conceptual model the relations between the main factors are illustrated. Although this
research limits itself to young adults, it is known that older people do not share the same safety
perceptions as younger people (Brå, 2014). This shows that age matters when looking at
characteristics of public spaces, regarding safety perception. Another example on how feelings of
safety are shaped by gender: in general men feel safer than women, and fears are not shared among
the two groups. Women have a greater fear of becoming a victim of crime in general compared to
men, yet statistically men are more often the victims of these crimes (Brå, 2014). People’s
occupation for a large part determines their movement through space. Women feel safer in cars in
public spaces, so a woman that uses the car to get to and from work has a different experience of
public space than someone that travels to work by bicycle or by foot (Valentine, 1990). Then again,
experience of being in public spaces only determines so much: someone who travels by foot through
a public space might be less scared of its characteristics than someone who travels by car through
the same area because they are not in direct contact with the area. This factor stands in relation to
the frequency with which individuals travel through public spaces.

Figure 2.1: Conceptual model by the author, based on literature from the theoretic framework
12
2.7.2 Perception public space
The characteristics of public spaces are shown in the top right of the conceptual model. With green
infrastructure, the characteristics of the area regarding green spaces are meant. The way green areas
can manifest themselves could have a direct influence on the perceived safety. It needs to be noted
here that the characteristics of public spaces in the top right square in the conceptual model,
influence each other as well; the amount of green in an area does not determine by its own the
degree of perceived safety, it is dependent on other factors such as presence of lighting
(Maruthaveeran & Konijnendijk van den Bosch, 2014).

The time of day determines for a large part the amount of people travelling through the city center:
some moments during the day are busier than others. Rush hour for example has many people
travelling from their homes to their workplace. When looking at a city center such as being done with
this study (the city center of Stockholm), the days itself could also potentially play a role; presumably
weekdays have a different ‘audience’ travelling through the city center, while in the weekends this
‘audience’ might be a bit more diverse, with shoppers and leisure seekers. Depending on other
factors mentioned in the conceptual model, such as the time of day, crowdedness could play a role in
the way safety is perceived by individuals. Large areas with lots of people could be perceived as safe
because the people in the area would keep an eye on each other (securing by community) (Mehta,
2014, p. 60). However someone else might find large areas with lots of people in them unsafe, due to
the possibility to become a victim of theft or other crimes, and would prefer smaller areas with fewer
people.
Within the existing literature, there is some controversy regarding the securing of public spaces, and
how people perceive this process. What becomes clear from the literature is that securing a space is
not as simple as it might sound. The presence of policemen patrolling an area to keep it safe might
appeal to some, and not to others. While some might find it comforting that the police are keeping
an eye on everyone, others might feel that the presence of policemen indicates that the area is not
safe and that this active patrolling is apparently needed (Mehta, 2014, p. 60; Brands et al., 2015).

The cleanliness of an area, together with the maintenance of green facilities plays a big part in how
the community treats the area. The maintenance factor within the conceptual model has its roots
from the broken window theory, and how a well-maintained area leads to fewer opportunities for
crimes or disorder. The broken window theory however does state that the opposite might also
occur; an area that is ill-maintained and has vandalized parts, makes it easier for offenders to act out
in the area (Gau et al., 2014). There is not a universal model for maintaining a public space and
expecting there to be no crime or disorder, but there are simple ways to reduce chances on crime
which come from the broken window theory. Well maintained green areas, such as flower gardens in
parks and the trimming grass fields, give people the idea that the government cares about the
esthetics of the area (Brands et al., 2015). By adding ‘maintenance of the area’ to the conceptual
model, this factor could give an insight in the value of young adults acknowledge towards esthetic
values and general perception of public spaces, and how they reflect on this process within safety
perception.

Other than the time of day that plays a role in the safety perception of an individual in public spaces,
seasonal changes also have an impact on the way people perceive space around them. People tend
to feel safer during the summer times, when there is more daylight which for a lot of people means it
is safe. During darker periods of the year, such as fall and winter season, people tend to not be as
active in public spaces (Maruthaveeran & Konijnendijk van den Bosch, 2014, p. 5). The seasonal
changes differ quite a lot from country to country (and thus from city to city). In Stockholm the
amount of daylight during the summer months (June, July and August) amounts to an average of

13
about eighteen hours per day. In the winter months (December, January and February) the amount
of daylight amounts to an average of about seven hours per day in Stockholm (Visit Sweden, 2015).
The difference between the winter- and summer months is quite big, therefore it is assumed that the
relevance of the effects of seasons on the perception of safety in public spaces, coming from the
literature, will have a similar impact on the safety perception among young adults in Stockholm (if
not more).

Related to seasons and (lack of) daylight is the availability of streetlights in public spaces. It would be
easy to assume that streetlights are always associated with having a positive effect on the perceived
safety of individuals in public spaces, however according to Lindgren & Nilsen (2011) this is not
always the case. The study does not deny the positive effects streetlights can have on safety
perception, instead Lindgren & Nilsen (2011) point out that streetlights alone don’t have very much
value, information wise. The combination of certain types of green facilities with streetlights can
have the opposite effect on an individual, and instead of feeling safe, the individual feels unsafe and
would rather avoid said area. Additionally the characteristics of the area overall play a more
important role in safety perception than just a singular factor which is ‘availability of streetlights’,
meaning that all of the factors mentioned under the ‘perception public space’ category together
form an image for the individual (Burgess et al., 1988; Mehta, 2014).
Public space is a concept that embraces many different locations with different characteristics. Even
within certain categories of public spaces a wide variety can be found; when someone mentions
‘parks’ as an example of public space, there are still a lot of questions to determine what the person
meant with ‘parks’, such as: what is the size of the park? Is a certain amount of green required for it
to be a park? Is it still considered to be a park when it allows light traffic to pass through the space?
These questions show that there is not a cookie cutter recipe for public space, and therefore the way
people perceive these varieties of public space needs to be treated carefully. The individual
characteristics of studied spaces need to be taken into account. The environment outside of the
public space can have an influence on the safety perception of the public space itself, meaning that
even though the public space itself has all the qualities to appear ‘safe’, if the environment
surrounding the public space does not supplement these qualities then people will feel less at ease
(Mehta, 2014, p. 60).

There is a difference between perceiving safety when being by one’s self and being in a group (for
example commuters sharing the same route). Preferences can differ from individual to individual, but
in general people prefer to travel through public spaces in the company of others than alone. This
moving through space can alter the way people perceive the space around them. If someone travels
through an area in a group, the person will most likely have less awareness of potential unsafe
characteristics of the area than someone who travels through an area by themselves (Brands et al.,
2015).

2.7.3 Degree of perceived safety

The abovementioned factors all have influence on the perceived safety, which is shown by the
arrows illustrated in the conceptual model (figure 2.1). The personal characteristics of an individual
(for a large part) determine the way someone sees the characteristics of public spaces, and the way
they feel about these characteristics. For example, women tend to recognize potential danger earlier
than men, and therefore tend to worry more about becoming a victim than men (Brå, 2014;
Maruthaveeran & Konijnendijk van den Bosch, 2014; Valentine, 1989). The characteristics of public
spaces in their turn have an effect on perceived safety, as they function as a sort of filter or
requirement list for an individual to either feel safe or unsafe. However, the individual characteristics
(such as age, gender etc.) also have a direct influence on safety perception; older people tend to feel

14
more unsafe than younger people. It must be noted that the relations drawn here are not static or
mutually exclusive; instead both the individual characteristics and characteristics of public spaces are
closely connected.

The outcomes from the theoretical framework chapter, resulting in the conceptual model, will be
used for formulating questions in the survey. The posing of relevant questions in the survey is of
importance regarding the hypotheses within this thesis, and eventually the answering of the main
research question. The discussing of the different hypotheses will be discussed in chapter 3
paragraph 3.2.

15
3. Methodology

3.1 Introduction
Within this chapter the method(s) applied in this thesis will be discussed, as well as the sub questions
that are created in order to eventually answer the main research question. Additionally, the target
group of this study and the research area are explained, so that a detailed image of the study is
given. The research area chosen for this study is also explained and illustrated. This chapter will
conclude itself by discussing the survey in detail.

3.2 Sub-questions & hypotheses

In order to be able to answer the main research question of this thesis: “in what ways does safety
perception affect the use of public spaces by young adults”, several sub questions are needed so that
different topics in this study are covered. However, the sub questions will not be answered directly.
In order to perform quantitative research, hypotheses are needed for statistical testing. A hypothesis
is a statement, based on previous studies or assumptions of the researcher, which links two (or
more) aspects within a study together, to see if there is a connection between the two aspects.
Usually statistical data is gathered regarding both variables in the hypotheses, and the appropriate
statistical tests (related to the variables) are applied in order to determine whether or not the
hypothesis is found to be significant. Judging from the outcome of the statistical testing, a hypothesis
is either accepted or rejected. The rejection of a hypothesis does not mean that the result of the
hypothesis is not interesting or wrong; however it does mean that no further scientific conclusions
can be made using that hypothesis as a source.

Below three sub-questions and their respective hypotheses can be found. Together the answers to
the hypotheses and the sub-questions will contribute to formulate an answer for the main research
question that will be answered in the concluding chapter of this thesis.

1. In what ways does safety perception differ between men and women?
- Women perceive public spaces as less safe than men
- Environmental factors (such as the presence of trees, bushes, water etc.) in an area are
more important to women than men regarding safety perception
2. Which factors seem to be the most important with reference to safety perception in general (by
both men and women)?
- Women experience night time as more threatening to their safety than day time
compared to men
- Different seasons have a direct influence on the safety perception for both men and
women
3. How can public spaces be improved in order to grow a positive sense of safety among the
visitors?

In what ways does safety perception differ between men and women?
The first sub-question targets the difference between men and women in regard to safety
perception. From the Swedish National Crime Survey (Brå, 2014) it becomes clear that women are
more afraid to become victims of crime than men. This sub-question offers a platform to confirm this
among young adults and to look at possible factors that influence this difference. The first hypothesis
covers that particular aspect, by stating that women’s safety is more influenced by crime perception
than it is for men. The second hypothesis brings more insight in the value of the presence of
environmental factors; perhaps both men and women among the young adult target population

16
don’t value environmental factors at all, which then has an impact on the way urban planning has its
influence on safety creation in public spaces.

Which factors seem to be the most important with reference to safety perception in general (by
both men and women)?
The second sub-question does not make a distinction between men and women and takes the
participation group as a whole. By asking the participants what they think are the most important
aspects about safety in public spaces, in the form of several multiple answer questions on the survey,
information about perceptions and experiences can be gained relatively easy. From the literature it
became clear that the lack of daylight can have a negative influence on people’s safety perception.
The winter months in Stockholm substantially deprive its inhabitants of daylight, which makes the
nights a lot longer than in the summer months. The first hypothesis for this sub question confirms or
denies the effect daylight has on safety perception among the target population for this research. For
this hypothesis women are taken separately to prove or disprove the significance of daylight. The
second hypothesis takes the influence of daylight on safety perception and takes it to the seasonal
level; if people feel less safe during night times, would they feel less safe during winter months than
in the summer months?

How can public spaces be improved in order to grow a positive sense of safety among the visitors?
The third sub-question takes the answers from the previous sub-questions and hypotheses, and
formulates goals to improve public spaces in regard to the perceived safety. The outcomes from the
previous hypotheses, together with the outcomes of the surveys, form a way for the third sub
question to be answered by analyzing the feedback given from the respondents in the surveys. One
section of the survey is dedicated to map questions, in order to offer respondents a way to voice
their feelings of being unsafe and to determine if there are areas that stand out as being specifically
unsafe. The map questions ask the respondents to fill in where they feel most unsafe during certain
times of the day or during different seasons. Additionally, the respondents are then asked why they
would feel unsafe in these particular areas. The marked areas by the respondents will result in
several heat maps, which show the most marked spots (the most unsafe areas) with red/yellow
colors and the least marked spots (the safer areas) as dark blue spots (figure 3.2). Using the spatial
data from the heat maps in combination with the reasons given why these particular areas are
perceived as unsafe will offer the respondents an easy way to state their feelings of safety
perception.

In addition to the in-depth statistical tests that are needed to accept or reject the hypotheses,
descriptive statistics will be discussed in detail as well. Descriptive statistics contain information
about the target group and will tell the reader more about the individuals that participated in the
survey, such as median/average age, the amount of men and women participating and other
relevant information regarding personal characteristics of the participants in this study.

3.3 Quantitative research: survey and statistics

The research question and topic usually dictate which kind of methods should be applied in order to
successfully complete a study. For the topic of this thesis, quantitative methods are chosen. From the
existing literature on this topic several important factors in regard to safety perception among
individuals manifest themselves. Using quantitative methods, in the form of a survey, a relatively
large amount of data can be collected and transformed into a database. The information stored in
this database can then be subjected to numerous statistical tests, in order to verify or reject the
hypotheses. The strength of quantitative research is that it gives the researcher the ability to make
statements about relations between different actors, by applying statistical testing on the

17
information gathered from existing databases or from surveys (‘t Hart et al., 2009). The quantitative
research for this thesis will be able to offer insight in the relation between individuals, how public
spaces are perceived and which factors contribute the most to feeling safe in public spaces. Using
relatively large amount of data gathered from the surveys in SPSS, generalizing statements can be
made regarding the population (not just about the respondents), if the respondents are a
representative part of the larger population (which needs to be tested and proven). The data from
the respondents, which will be taken into the statistical testing procedure for this thesis, is focused
on young adults.

3.4 Target group

The target group for this thesis is young adults. From the literature and available statistics it becomes
clear that there is a difference between younger and older people regarding safety perception. Older
people tend to perceive areas and their own situation as less safe compared to younger people (Brå,
2014; Maruthaveeran & Konijnendijk van den Bosch, 2014, p.3; Mehta, 2014, pp. 81-82). This thesis
will look into further into what the thought process is amongst younger people which makes them
feel safe, and if there are differences in perceptions among young women and men in safety
perception. For this thesis, young adults are considered to be between the age of 18 and 30 years
old.
For this thesis, a distinction and a definition of the concept ‘young adults’ is needed, due to
the representativeness that is required in order to make valid statements about the gathered data
from the survey. Representativeness is essential in quantitative research because it allows the
researcher to make statements that are not limited to the respondents of the study, but the larger
population the respondents represent (‘t Hart et al., 2009, p. 108). This means that for this study,
statements regarding young adults in general (if found representative) can be made, instead of being
limited to making statements about the relatively small group of respondents to the survey. The
concept for this process is called ‘external validity’; making sure the population is validated
(representative) which will determine the strength of the statements made further on in the analysis
chapter of this thesis. The population validity will be tested using information gathered from the
third and last part of the survey. This part of the survey contains information about the respondents,
while maintaining anonymity. Using the answers on the basic questions regarding gender and
achieved educational level, and comparing them to the statistical database of Sweden (Statistiska
Centralbyrån/Statistics Sweden, 2015), the gathered information of the respondents can be
validated.

3.5 Research area

The outcomes of the theoretical framework, concluding with the conceptual model, dictate the
factors that influence the safety perception of an individual in public spaces. Aside from using this
conceptual model as the basis for the quantitative research, the outcomes from the theoretical
chapter will also be used to choose a research area. Because of the many environmental and spatial
factors that influence safety perception, choosing the right research area is vital for the thesis. The
area that has been selected is what could be considered the city center of Stockholm. The area has
the size of four squared kilometers and an outline of 8,36 kilometers. Inside the boundaries of the
selected area the central station, the relatively crowded main square of Stockholm at Kulturhuset,
Skeppsholmen, Kungsträdgården, a part of Djurgården, and Gamla Stan can be found (figure 3.1).

18
Figure 3.1: Selected research area: City center of Stockholm, Kungsträdgården, Skeppsholmen, part of Östermalm, part of
Djurgården, and Gamla Stan (Google Maps, 2015)

The area containing the central station and the crowded main square are both characterized by big
open spaces, underground passages and big streets. Additionally the presence of unknown
individuals in these areas could pose a threat on the safety perception, during both day- and night
time. Skeppsholmen is a small peninsula on which several museums and offices are located. It is
expected that this area will have fewer visitors than Djurgården, mainly because of the geographical
location of Skeppsholmen and the limited services it has to offer. Djurgården is a relatively green
peninsula with lots of trees, bushes, grass fields and water. The area contains a lot of museums and is
rather popular in the summer among tourists and the Stockholm inhabitants themselves.
Djurgården and Skeppsholmen could be seen as connected with each other through a ferry that
travels from the south of Gamla Stan to the two peninsulas. Kungsträdgården is an area which is
located east of the main square. It is officially marked as a public park, yet it also has the
characteristics of a square, because of its open spaces, relatively few trees and bushes normally
associated with a park and geographical shape (Kungstradgarden.se, 2015). Kungsträdgården is well
connected with the rest of the city, and could very well be marked as the most center located park in
Stockholm. The park holds many events, especially during the summer. Gamla Stan is the old city
center of Stockholm, which is nowadays mainly focused on tourists. The main streets of Gamla Stan
contain many cafés, restaurants and tourist merchandise shops. The area has small streets and most
of them are ill-lit.

The selected research area fits the environmental factors described in the conceptual model, in order
to facilitate the optimal way of asking respondents about their perception in said factors (figure 2.1,
page 13). Even though the research area fits all the criteria found relevant in the literature (see the
Theoretical Framework chapter, page 12), the characteristics of the very center of Stockholm city
need to be stressed. It is an area that contains lots of water, green spaces, and is also very urbanized.
These characteristics determine the ‘ecological validity’ of this research; the environmental

19
characteristics of the chosen research area must be taken into consideration if this research is copied
and used into a different setting (another city or region) (‘t Hart et al., 2009, p. 153-154).

3.6 Heat maps

For this research several map questions within the survey are needed. In order to make this task
operational, respondents will be given the option to mark a maximum of three areas (which could be
a street, a square and a part of a park) that are experienced as the least safe. The importance of
daylight and how it affects people’s safety perception is taken into account in the survey, and
therefore the map pinpoint question will be asked for both winter and summertime. The reason to
offer the respondents more than just one map is due to the sensitive nature of daylight and seasons
that have an influence on safety perception (Bennett et al., 2007; Shenassa et al., 2006). In addition
to the map pinpoint questions, which are rather specific, respondents are also asked to mark an area
that in general they experience as least safe. This is done by splitting up the research area in different
zones. Some areas within the research area are quite large, but perhaps should not be split up. For
example Gamla Stan is a relatively big area, but the area itself is quite homogeneous and therefore is
treated as one ‘zone’ (figure 3.1). The area near the central station of Stockholm is also quite big, but
has more diversity and is therefore split up into four zones, to give people the chance to be more
specific.

Figure 3.2: Example of a heat map with several spots marked as unsafe (Google maps,
2015; Qualtrics, 2015)

For the survey, the research area will be presented to the respondents in a map from Google Maps.
This is done to make sure that the respondents have to spend relatively little time getting to know
the map and the selected research area, because Google maps is one of the most well-known map
distributors. The results from the map related questions in the survey will result into ‘heat maps’.
These are maps that show information using graduate colors. The brighter areas marked on the map
are experienced as unsafe, while the darker areas marked on the map are experienced as safe by the
respondents (figure 3.2). Heat maps, or choropleth maps, are a way to visualize the outcomes of
spatial data. The spatial data within this thesis are the locations that respondents pinpoint in relation
to feeling unsafe during specific seasons and times of day.

20
3.7 The survey in detail
When using a survey as the main method of gathering empirical data, the way questions are posed
can determine the response rate for a large part. Additionally, if the survey contains questions that
are difficult to understand for the respondents, the chance to get a lower response rate is also
increased. The use of jargon in surveys (such as difficult terminology for which you need some sort of
background information in order to understand it) is often not appreciated, depending on the
audience you’re trying to reach for response (‘t Hart et al., 2009). Questions, or for that matter entire
surveys, that people cannot understand will be either skipped or filled in wrongly. The filling in of
questions wrongly, or non-true, by respondents changes the usability of the data overall.
For this thesis, the survey is spread using two main methods: using online, social media
methods and in-field surveying. When doing in-field surveying the researcher usually asks potential
respondents if they have time to fill in a survey, after briefly explaining what the actual research is
about and how the filled in surveys will be treated with strict confidentiality and anonymity (‘t Hart
et al., 2009). Surveying over the internet has more flaws than in-field surveying; people can more
easily decline filling in a survey because they are not directly confronted with it. Additionally, over
the internet you cannot tell if people are who they claim to be in the survey. The internet forms a
barrier which prevents the researcher from seeing if the respondents fit the limitations set out by the
researcher, for example being in the right age group (in the case of this thesis, being between
eighteen and thirty years old). However, if used right, the internet can also act as a snowball,
meaning that if respondents are interested in your research, they could be inclined to share the
survey with other people, from which the so called snowball effect arises (‘t Hart et al., 2009). Both
using the Internet and in-field surveying as ways of gathering data using the survey method have
their advantages and disadvantages, and both should be labeled differently to ensure that
representativeness and the non-response rate can be calculated and determined properly.

Below, the survey used to gain empirical data will be discussed in detail, including the decision to
include especially these questions. The survey was created on a website from the company
‘Qualtrics’, because it allowed the use of map related questions which added an interactive platform
for respondents to voice their opinion regarding feeling safe in public spaces. The data gained from
in-field reporting was later copied into digital surveys, to make map analysis possible.

3.7.1 General sense of safety – Part 1/3 of the survey

The first part of the survey is supposed to ask relatively easy-to-answer questions, to give the
respondents a basic idea of what the research is about. By asking a combination of ‘yes or no’ and
multiple choice questions regarding feeling safe in public spaces, the respondents were introduced to
the topic, yet a distinction between respondents who sometimes feel unsafe and respondents that
never feel unsafe can already be determined. The respondents who filled in that they never visit
Stockholm were routed to another question asking the selected respondents why they never visit
Stockholm. This question is not of great value on its own, but one of the possible answers among the
multiple choice answers was “because I feel the city is too unsafe for me”, which could be a good
reason for people to avoid the city. However, it must be noted here that it is expected that only very
few respondents will fill in that they never visit Stockholm, as the surveying is focused on inhabitants
of Stockholm and the Stockholm area (suburbs).
The respondents are then asked about different factors that play a role in perceived safety.
The factors given to the respondents stem from the theoretical framework chapter of this thesis, and
can also be found in the conceptual model (figure 2.1). This question has an interactive aspect to it:
respondents are asked to drag and drop the factors in specific orders, from 1 (most important) to 7
(least important). The potential downside of this question is that there already is a default order,

21
which means that respondents could more easily agree with the default order than their own actual
opinion.
The following two questions in part one of the survey contain questions that are answered by
all respondents, regardless if they have answered that they do not feel afraid in public spaces or that
they have not visited Stockholm city. These are also the last two questions of part one of the survey,
which means they still focus on the general sense of feeling unsafe and public spaces, without going
into much detail such as looking at the research area. The sixth question in the survey is using the
likert-scale method. This means that a certain statement is posed to the respondent, and it is up to
the respondent to fill in to what degree they agree or disagree with said statement. The strength of
these likert-scale questions is that as a researcher, several hypothetical situations can be stated.
Subsequently, respondents then can choose from five possible answers they can give to these
statements or situations: totally disagree, disagree, neutral, agree or totally agree. For example
within the survey, some general statements were made such as: “most public areas in Stockholm are
unsafe” and “I tend to avoid small/narrow streets when it is dark outside”. These statements are not
too specified to a certain situation, and give the respondent a relatively clear option when it comes
down to answering the question; as a respondent you either agree or disagree with these statements
(most of the time). The agreeing and disagreeing on likert scale questions is rather important. Within
the given options for answers, a “neutral” answer must always be presented to the respondents,
because it is possible to have no affiliation with a statement or that it simply does not apply to the
respondent’s situation. However, having a neutral answer within the given answers can pose a threat
to getting good, meaningful results. If a lot of respondents fill in that they feel neutral towards most
statements in the likert-scale questions, it becomes harder for the researcher to use the data for
statistical analysis and therefore to draw conclusions (other than: most people feel neutral towards
these and these statements) (de Vocht, 2010). Therefore, to gain the most usable results for
statistical analysis, the researcher has to pose statements in a way to challenge the opinions of
respondents. This can be done by writing a statement that is expected to be answered negatively
(‘totally disagree’ or ‘disagree’) in a positive way. In a previous example the statement “most public
areas in Stockholm are unsafe” was used. This is also an example of a challenging statement which is
expected to get respondents to avoid the ‘neutral’ answer, and instead agree or disagree with the
statement. The second statement in question six of the survey does not have the factor that it
appeals to everyone; the respondents that filled in that they did not feel unsafe in public spaces most
likely will not have a strong opinion on this statement. However, in relation to the outcomes of the
literature regarding crowdedness and travelling within a group (of commuters or friends) through
public spaces and how this affects one’s perception of self-safety, the answers given to this
statement could provide some degree of insight to this topic. The statement is phrased as such: “for
my own safety, I would rather walk through Stockholm with a friend /colleague than alone”. The first
part of the statement (“for my own safety”) is added to make sure people would not confuse walking
with a friend purely for entertainment purposes, instead of safety purposes.
The last two statements of question six are less situational, and could be applied to different
contexts as well. Asking respondents about how they feel regarding public spaces in different
settings, such as different seasons, gives an insight in what influences (the lack of) daylight has on
safety perception. The potential link between a change of seasons and shifts in safety perception is
further questioned in the last question of part two of the survey. Three statements are presented to
the respondents, similar to the previous question. One of the statements offers another option for
respondents to agree or disagree on, which is the weather during different seasons. The statement
“during the winter it is not the darkness that makes me feel unsafe, it is the weather” puts the
difference in daylight during the winter and summer aside, and instead discusses the possibility that
for example snow, or warm sunny weather, has a stronger influence on safety perception. These last
three statements, which form question seven in the survey, are a bit more complicated and

22
challenge the respondent to think about their own situation and how they would feel. This builds up
to the more complicated questions in part two of the survey.

3.7.2 Feeling unsafe in the city – Part 2/3 of the survey

This second part of the survey characterizes itself with a lot of interactive elements built in the
survey. The respondents are asked to identify several spots that they experience, or have
experienced, as unsafe. The data that is generated from these map related questions do not contain
any GPS coordinates; however they do offer data visualization in the form of heat maps. These heat
maps can be used to identify certain hot spots when it comes down to unsafe public places; if a lot of
respondents mark a particular area as unsafe, it could be interesting to look into the reasons why
people experience this place as unsafe. Therefore respondents are also asked to give a reason why
they feel that the selected areas are unsafe.
The first question of the second part starts off relatively easy, by asking the respondents to
identify a particular area within the research area (which is a part of the city centre of Stockholm)
that they feel is unsafe in general. The follow up question to this basic question is an ‘open question’,
which asks why they selected the area in the previous question as unsafe. This means that
respondents can formulate themselves why they’ve chosen this area. The choice for an open
question here is because limiting the answers by giving the respondents a multiple-choice question,
with answers based on findings from the literature, could prevent respondents from sharing a more
valuable experience within the area.
The second map related question asks respondents to mark a maximum of three points in
the research area that they would avoid after 20.00 o’clock during the winter. With this question it is
important to stress the situation posed to the respondents, because if this is not clear to the reader
of the survey, then the reader could possibly consider other areas to be unsafe, or to feel that none
of the areas are unsafe. Other than the first map question, the second map follow up question is a
multiple-choice question, but leaves room for respondents to fill in their own answer if they are not
satisfied with the answers offered by the researcher. The possible answers given at this multiple-
choice question are derived from the literature and can be found within the conceptual model as
well (figure 2.1).
The third and last map related question of the survey is the same as the second one except
the situation is changed. This time respondents are asked to mark a maximum of three points in the
research area that they would avoid at 20.00 o’clock during the summer. The goal of these two
questions (both the winter and the summer situation) is to see if there is a significant difference
between the two seasons and the way people perceive areas as safe or unsafe. The same multiple-
choice follow up question is given to the respondents, which concludes the second part of the
survey.

3.7.3 Personal characteristics – Part 3/3 of the survey

The last part of the survey consists of relatively basic personal questions about the respondent.
Within the third part of the survey the respondent is asked to answer questions about gender, age,
occupation, highest degree obtained etc.. The data from these questions is used for the obvious
comparison between men and women and how both groups perceive personal safety in public
spaces (differently). The first question in the third part asks respondents to fill in their gender. The
data generated from this question on its own is not that interesting; it only shows the amount of
men and women who filled in the survey. However the main goal of this question is to be able to
relate the data with data from other questions in the survey. In order to determine if there is a
significant difference between men and women and the value of streetlights in relation to public
safety, combining outcomes from several questions is needed. Another example is to test the
amount of highly educated people versus lower educated people in regards to public safety; do

23
higher educated people feel safer than lower educated people? Or perhaps it is the other way
around?
The second question in the third part asks respondents to fill in the year in which they were
born. Instead of asking respondents directly how old they are, a question that allows respondents to
only fill in the year in which they were born distances itself from becoming ‘too personal’ (‘t Hart et
al., 2009; Chicago manual of Style, 2010). The following questions ask respondents about the highest
obtained degree and their occupation. This study focuses on young adults, of whom a significant
amount is student, and therefore there is not much variation in occupation. With the data from these
two questions and the former questions the representativeness of the respondents can be tested.
The representativeness of the respondents determines if any statistic statements can be made and if
so, in what way they represent the rest of the young adults living in or near Stockholm city.
In the third part of the survey respondents are also asked if they were born in Sweden or
elsewhere. These two questions don’t have the goal to try and find a difference between natives and
migrants regarding safety in public spaces, but the division of natives and migrants among the
respondents also determines the representativeness. If people were not born in Sweden they are
asked to answer a multiple-choice plus-own-answer question in order to determine the country in
which they were born.
Following the questions about the country of origin is the question that asks respondents to
fill in how long they have been living in or near Stockholm. This question is a multiple-choice answer
style question with four options: ‘less than a year’, ‘between 1-5 years’, ‘between 5-10 years’ and
‘longer than 10 years’. This question was added to the survey in order to determine if there is a
significant relation between the amount of time lived in or near Stockholm, and the experience of
public safety.

At the end of the survey respondents have the opportunity to give any comments of feedback on the
survey or on the topic in general. An option has been added to the survey for respondents to fill in
their email address in order to receive a PDF copy of the thesis when it is finished.

3.8 Ethical issues

The ethical issues one could encounter while writing an academic piece (thesis, dissertation, paper or
essay) depend on the topic and the research methods used by the researcher. Within this thesis
quantitative methods are used, which are quite different than qualitative methods such as interviews
or focus groups. When interviewing, one has to pay attention to what impact the researcher can
have on the interviewee, which in its turn could affect the outcomes of the interview and the study
itself. This is not so much the case for quantitative methods, although there are definitely some
important potential ethical issues the researcher needs to take into account while using quantitative
methods. When using a survey as the main method for gathering empirical data, the structure and
the way questions are posed can affect the response rate and therefore the results of the study. In
the case of this study questions are asked about personal information at the end of the survey, such
as the age, gender and highest achieved educational level of the respondents. These are questions
which if asked in a wrong way, could refrain respondents from willing to participate or fill in crucial
data. When asking about the age of a respondent it is considered better to ask about the year in
which they were born, instead of asking directly how old they are, the former feels less like a direct
question and therefore makes it easier for (some) respondents to fill in said question. However, this
thesis focuses on young adults, meaning that perhaps this issue is not that relevant within this study.
Another ethical issue that comes into play when using quantitative methods is anonymity
and secrecy. When giving people a survey to fill in, the researcher expects a lot from the respondent;
respondents share their feelings regarding a certain topic and are asked to fill in some personal
information as well. It is the job of the researcher to state clearly that the information gathered from

24
the surveys is treated strictly confidential and anonymously, in order to put the mind of the
respondents at ease. The survey from this study informed respondents about the intentions of the
study and how their data would be handled. This information was repeated once more when
respondents had to fill in some personal data (gender, age and household income), to assure the
respondents that their data would be handled with care.
Additionally, the respondents were offered to fill in their email address at the end of the survey to
receive the final version of the thesis in PDF format.

3.9 The process of surveying

The experience of surveying in the research field within Stockholm city resulted in different response
rates from day to day. On days that had great weather with a lot of sun, people were less hesitant of
being approached, and were more eager to fill in the survey. Rainy and cloudy days yielded less
response compared to sunny days. In hindsight this is good to know, which could help with future
surveying. One of the possible factors could be that on sunny days people tend to be happier and
enjoy being outdoors, which helps when asking people if they could help you out by filling in a
survey.
The spreading of surveys online was done at two timeslots. The first wave with which the survey was
spread out was done at the beginning of the gathering of empirical data phase. The second wave was
done near the end of the allotted gathering of empirical data timeframe. The reason for a second
wave of surveys was done in order to catch potential respondents that missed the first wave.
However, before sending out the surveys to online platforms, a selection of possible target groups
(websites) was made in order to control the spreading of the survey. Several websites with discussion
forums that would potentially welcome the subject of this research were targeted; thelocal.se,
reddit.com/Sweden and a Facebook group for people living near Stockholm University and the city
centre of Stockholm. These websites attract a diverse group of people; people working in Stockholm,
people looking for jobs in Sweden and Stockholm, and people studying in Stockholm. Because of the
audience found on the selected websites, the respondents had a relatively high chance of being in
the selected age group (18 years old to 30 years old) set out for this study.
After collecting the data from both the online survey and the surveys conducted in the field,
the amount of online surveys and in-field surveys were compared. After comparison it appeared that
both methods yielded about the same amount of results; the online version yielded 64 filled in
surveys and the in-field surveying yielded 43 filled in surveys. It is important to note here that the
amount of surveys yielded per ‘method’ can differ, dependent on several factors: (1) the time of day
in which the researcher both sends out the surveys online and does in-field surveying, (2) the
message sent out online which tries to get people’s attention and willingness to cooperate and the
brief presentation of the topic when surveying in the field, (3) the weather when surveying in the
field and (4) the impact of being approached by a male researcher. Additionally, not every survey
that was collected could be used for research, due to respondents not fitting the target group.

The time of day in which the researcher sends out the survey on selected websites and does the in-
field surveying has a big impact on potential results. People usually have time to browse websites
during lunch breaks or after work (of course there are exceptions). Additionally, the weekend is an
important factor to take into account regarding the spread of the survey via websites; even though
work is not a factor in general during the weekend, perhaps people spend more time outside or
doing other activities than sitting in front of the computer. The same principle goes for surveying in
the field during certain times of day; it could be beneficial to survey during rush hours, because this
would technically yield the biggest potential response group. However, rush hour is often associated
with people being in a hurry to get from A to B, meaning that perhaps the people travelling during
rush hour do not want to take the time to fill out a survey.

25
The second factor that had influence during this research is the way the survey is presented to
people. When using a short and catchy message, people tend to get curious which could lead to
potential respondents. It is therefore of importance to come up with a good presentation regarding
the topic of the survey and the research, both online and for in-field surveying.
The third factor that had influence on the gathering of empirical data for this study was the
weather. This factor affected only the in-field surveying, as more people were more likely to
cooperate when it was sunny outside than when it was gray and/or raining. This is a relatively
straightforward factor as one could imagine walking through the rain; it is not really attractive to stop
and take the time to fill in several papers with questions.
The fourth factor relates to the gender of the researcher in relation to getting potential results. Even
though the topic of this study is public safety, among other subjects covering the topic of ‘fear of the
other’, this gender factor could be applied to all in-field survey studies. The gender of the researcher
approaching people on the street could affect the willingness of people to stop and listen to what the
researcher has to say.

The before mentioned factors affect the empirical data yield. Other research methods, which use a
more qualitative approach, have similar factors that could affect the empirical data. However, in the
case of interviewing people, it is more likely to affect the quality of the data than the quantity of the
data. A male researcher interviewing a woman about a sensitive topic has a chance to get less
complete or honest answers than a female researcher conducting the same interview for example
(hypothetically speaking).

26
4. Representativeness and descriptive statistics

This chapter will discuss the first results from the empirical data that was gathered during the study.
In order to be able to make any statements about any significant relations between two or more
factors regarding public safety, the representativeness of the empirical data needs to be tested and
confirmed. The representativeness of the gathered data will determine in what ways statements
about the hypotheses in this study can be made. This is done in order to be able to make statements
not only of the respondents that helped with this study, but also about the overall young adult
population living in or near Stockholm city.

As stated before, the target group of this study is young adults between the age of 18 and 30. In
order to calculate how large the sample size has to be for this study, it is needed to determine how
many young adults within the same age group as the study are living in or near Stockholm city. The
sample size of the survey determines approximately how many surveys need to be sent out in order
to gain enough results to be able to make statistical statements, thus accounting for non-response as
well. The total sample size of the survey is not the final amount of response needed, because not
everyone fills in the survey (due to lack of time, no interest in the topic, or just does not see the
survey), which is called non-response. The calculating of non-response depends on the circumstances
with which the empirical data is gathered, and will be discussed within this chapter in paragraphs 4.1
and 4.2.
The Swedish bureau for statistics (SCB) offers numbers regarding the population of all the
Swedish municipalities. In order to determine the amount of young adults that could potentially visit
Stockholm city, it is needed to look beyond the municipality of Stockholm alone. This is why sixteen
municipalities around the Stockholm city centre have been taken into account for this research. The
amount of young adults (between the age of 18 -30) living in the municipalities of Upplands Väsby,
Vallentuna, Järfälla, Ekerö, Huddinge, Tyresö, Täby, Danderyd, Sollentuna, Stockholm, Södertälje,
Nacka, Sundbyberg, Solna, Lidingö & Sigtuna come to a total of 325.000 young adults in the year
2014 (SCB, 2015). However this is not all the information we need in order to calculate the sample
size. The ‘strength’ with which statistical statements can be made needs to be determined here as
well. There are several options to choose from: a 90% reliability level with a 10% error margin, a 95%
reliability level with a 5% error margin and a 99% reliability level with a 1% error margin. Obviously
the latter option gives the strongest reliability when it comes down to testing several variables from
the survey. However the most common reliability level to pick (especially when it comes down to
calculating the sample size of the survey) is 95% with a 5% error margin, which is the reliability level
chosen for this study in order to calculate the sample size (Vocht, de, 2010; ‘t Hart et al., 2009).
It must be noted here that different reliability levels can be applied to statistical testing in the
next chapter, meaning that if two variables appear to have a significant positive or negative relation
with a 95% reliability level, it might be interesting to further test if these two variables still hold this
relation with a 99% reliability level.

4.1 Calculating the sample size and the minimum of responses needed
As determined earlier, the total population of which this study wants to make statistical statements
about amounts to 325.000 young adults living in or near the Stockholm city area. This is a relatively
large group of people; however this does not necessarily mean that the sample size of the study
needs to be equally as big. When the population of a study is larger than 20.000 the outcome of the
sample size formula does not change, meaning that the population for this study of 325.000 people
gives the same sample size as if it were 20.000 or 60.000 people.
There are several ways of calculating the sample size; one way is to determine which
reliability levels one wants to use and using those variables into the formula. Within this research

27
95% reliability is chosen with a 5% (otherwise written as α=0,05) error margin. These numbers
translate into other values when working with sample size formulas; 95% reliability with α=0,05 error
margin results in a critical Z-value of 1,96. The critical Z-value calculation is done using a conversion
table which is determined by the error margin paired with the reliability level. An error margin of 5%
or α=0,05 dictates that the Z-value becomes 1,96.

The formula used to calculate the sample size is:

n = minimum sample size
Z = critical Z-value with the chosen reliability level
E = desired accuracy (also known as the error margin, which is 0,05 in this case)

When filling in the variables that apply for this research the following formula shows that the
minimal sample size of this study is 384. This means that while accounting for non-response, the
amount of surveys that needs to be sent out is 384. However, with this study the surveys are not sent
out to people’s homes, instead people are asked on the streets if they have time to fill in the survey.
Additionally, the survey is spread online in various forums and Facebook groups, which are not
similar to the standard non-response rates and the sample size calculated. Therefore, it is needed to
look at different possible non-response factors that could affect the total response yield.
There are several similarities between ‘standard’ non-response rates and the non-response
rates relevant in this study. For example, posting a survey to several households in a certain
neighborhood does not result in a 100% response rate. Only a small group of people is willing to take
the time to fill in a survey and return it to the researcher. This same process can be seen when the
survey is spread online; only a small group of people is willing to click on a link and fill out a survey. It
is therefore important to make sure the researcher brings the message of the survey across in as few
sentences as possible (‘t Hart et al., 2009). This way the respondent can get an idea of what the
researcher’s intentions are and what the survey is about. When surveying in the field, in this case the
research area of Stockholm city chosen for this study, instead of given people a brief summary of the
study on paper for them to read, it is often a better choice to verbally communicate the intentions of
the study in one or two sentences (‘t Hart et al., 2009). Another difference between posting a survey
in paper form to households and posting it online (on a forum or Facebook group) is the frequency
with which results come in. When posting a survey online, usually the message containing the survey
(and a brief explanation of the study) is only viewed one or two days, or even only a few hours,
before other posts ‘bury’ the survey post. This means that there is a relatively small window for the
researcher to gain response. A survey on paper mailed to several households can take much longer
to be filled in. Additionally the time between the respondent filling in the survey on paper and the
researcher collecting the filled in survey can be much longer than respondents filling in an online
survey, which after filling is automatically collected by the researcher.

The minimum amount of surveys needed to be collected is determined to be 100. This number has
been chosen after taking the sample size of the survey into account, determining the factors that
could influence non-response and the timeframe within which the thesis needed to be completed.
The minimum of 100 responses is needed in order to make statistical statements about the larger
population, which are young adults living in or near Stockholm city. The minimum of empirical data
set also allows for internal testing within several groups (men and women, people who do not feel
unsafe vs. people that feel unsafe, people that value environmental factors more vs. people that
value the presence of other people more etc.) (‘t Hart et al., 2009). Within this study the survey has

28
been spread using two methods: surveying in the field and spreading the survey online on forums.
Because of these two methods, two separate non-response rates have been determined. While
surveying in the field there were mainly four responses from respondents when asked if they were
willing to cooperate with a survey regarding public safety: some people kept on walking or said that
they did not have time (1), other people stopped to listen to what the study was about but
afterwards said they did not have time (2), another group was willing to participate and fill out the
survey completely (3), the last group of people participated with filling in the survey but failed to
complete the survey (4).

The same survey online yielded somewhat similar results as the surveys that were conducted in the
field, regarding the way respondents felt about the time it takes to fill it in. Every response is
captured, meaning that if someone opened the survey in their Internet browser and closed it after
one minute and answering maybe one or two questions, the answers are still captured. However, the
validity of these types of responses could be questioned. Perhaps the respondent was just interested
in how long the survey was and found out it was too long and decided to not cooperate. The
responses that captured at least 90% of the required answers were taken into account. This is done
in regards to the validity of the research and the consent of the respondents and their personal
answers. By abiding to the 90% rule, the respondents that decided they did not want to be part of
the research and closed the survey before answering at least 90% of the required questions, it is
made sure that only respondents that were willing to participate and share their information are
used for this study.

4.2 Descriptive statistics: characteristics of the study

This paragraph will look into a specific part of the gathered data of the survey. By looking at the
amount of men and women, the ethnicity and the average age of the people that have filled in the
survey, a comparison can be made between the empirical data from this study and the statistics from
the Swedish Bureau of Statistics (SCB). The reason why this is done is to make sure that the gathered
data is representative for the larger population it represents. For example, if the balance between
men and women is too far off from reality, weights need to be added to the variables in order to be
able to make valid statements. Additionally, this paragraph will give some more detailed information
about the respondents regarding their perception of own safety in public spaces and living in or
around Stockholm city.

It must be noted here that with the variables tested in this paragraph, the term “population
Stockholm” will be used. For this study, this term refers to the municipalities closest to the city
centre of Stockholm. These municipalities are: Upplands Väsby, Vallentuna, Järfälla, Ekerö, Huddinge,
Tyresö, Täby, Danderyd, Sollentuna, Stockholm, Södertälje, Nacka, Sundbyberg, Solna, Lidingö &
Sigtuna.

A total of 103 responses were recorded and deemed usable for this study. The selection of usable
surveys was done by looking at the age of the respondents, if the respondent did not fit the 18-30
years old age group the response was not taken into account regarding the statistical analysis.
Additionally, respondents that failed to fill in at least 90% of the questions on the survey were not
taken into account either.

29
4.2.1 Gender
The amount of men and women that have participated with the survey differ somewhat to the
statistics of the population given by SCB4 (2015). The amount of male respondents amounts to 52,4%
and the amount of female respondents amounts to 47,6. The gender distribution is more equal when
looking at the statistics of the population of Stockholm, with both groups being almost equally
represented (figure 4.1). The calculation of the representativeness of the category gender in relation
to the statistical database of the population of Stockholm (SCB) will be determined in paragraph 4.3.
Figure 4.1 Gender distribution respondents compared to target population

Gender distribution
100%
90%
80%
47,6 50,2
70%
60%
Female
50%
Male
40%
30%
52,4 49,8
20%
10%
0%
Respondents survey Population Stockholm

Source: SCB (2015) & Own research (2015)

4.2.2 Age
The average age of the respondents is quite similar to that of the population of Stockholm (figure
4.2) (SCB3, 2015). The average age of the respondents is 24,8 years old. The average age of the young
adult population of Stockholm is 24,6 years old. It is not common to talk about age and to use a
decimal number, however when calculating the representativeness it is needed. The
representativeness of the variable age will be discussed in paragraph 4.3.

Figure 4.2 Age distribution respondents compared to target population

Age distribution respondents Age distribution population

Stockholm
18-21 18-21
years old years old
27-30 17,82% 19,65%
27-30
years old years old
36,63% 41,46%
22-26 22-26
years old years old
45,54% 38,88%

30
Source: SCB (2015) & Own research (2015)
The age groups created for this study were needed in order to determine the representativeness. By
looking at the average ages of the two groups it is obvious that the two groups are very similar,
however for statistical determination if the two groups are in fact similar, division in age groups is
needed. The way the different ages are divided into groups does not matter, as long as both the
respondent group and the population of Stockholm group have the same division (18-21 years old,
22-26 years old, 27-30 years old in this case).

4.2.3 Ethnicity
The ethnicity of the respondents differs somewhat from the population in Stockholm. This could be
because of the groups chosen within the Swedish Statistical Database (SCB2, 2015), using the
statistics regarding ethnicity of the group 15 to 34 year olds. This is a group that takes more ages into
account than the group of the survey (18 to 30 year olds). It is therefore expected that there are
some deviations between the respondents and the population of Stockholm. Using these statistics, it
can be determined if the respondent group is still representative regarding ethnicity or not.
The amount of people with the Swedish nationality is higher within the respondent group than in the
population Stockholm group (figure 4.3). This automatically lowers the amount of migrants in the
respondent group, which could affect the representativeness of the ethnicity variable.

Figure 4.3 Ethnicity respondents and population comparison

Ethnicity respondents and population

100%
90%
27,18
80% 36,02
70%
60%
Migrants
50%
40% Swedes
72,82
30% 63,98
20%
10%
0%
Respondents survey Population Stockholm (15-34 years old)

Source: SCB (2015) & Own research (2015)

31
4.2.4 Amount of time lived in or near Stockholm
One of the last questions asked to the respondents was regarding the amount of time lived in or near
Stockholm. The data gathered from this question could form the basis for further analysis between
respondents that have stayed in or near Stockholm for a short period of time (up until now) and
respondents who have lived for a longer period of time in or near Stockholm.

It becomes clear that a large part of the Figure 4.4 Amount of time lived in or near Stockholm by the
respondents to the survey has lived longer respondents
than ten years in or near Stockholm. The
smallest group among the respondents has Respondents: Time lived in
lived for a period of five to ten years in or or near Stockholm
around Stockholm. The fact that a lot of
respondents have lived in or near Stockholm
for longer than then years, and that relatively Less than
very few people have lived in Stockholm for a year:
12,6%
five to ten years indicates quite some diversity
among the respondents. This, in turn could Longer
result in different opinions regarding public than 10 Between
safety among the respondents, if the time years: 1-5 years:
lived in Stockholm has any effect on safety 47,6% 28,2%
perception. It must be noted here that the
Between
respondents themselves had the freedom to
5-10
decide what they define as living “in or near years:
Stockholm”, there were no boundaries given 11,7%
or a map of which the respondents had to
choose from. This has been done in order to
give more freedom to the respondent in relation to thinking about safety perception.

4.2.5 Employment status of respondents

The majority of the respondents within this survey are students (figure 4.5). This outcome is not very
surprising as the target group for this study is between the age of eighteen and thirty years old,
which is a common age to be studying. The second largest group, following the students, is the group
that is employed full time. The balance between the two largest groups, with students representing
about 53% of the respondents and the full time job group representing with about 38%, is quite
even. The other 9% consists of people either looking for a job, or doing volunteer work and having a
part time job.

32
 Figure 4.5 Current employment status of respondents

Respondents: Employment
status

Full time
job:
Student: 37,9%
53,4%

Out of
work and Part
looking Volunteer time
for work: work: job:
3,9% 1,0% 3,9%

4.2.6 Education level Figure 4.6 Education level respondents

Among the respondents there is a relatively even
distribution regarding highest degree obtained. The Respondents: Education
respondents were asked to fill in which level they had
obtained, resulting in the Bachelor’s degree being the
level
biggest group (figure 4.6). Looking at the target group Grund-
for this study, being eighteen to thirty year olds, it is Master's
skola:
to be expected that there are so few people who have 2,9%
degree:
completed grundskola as the highest obtained 25,2%
degree. The second biggest group consists of the
people that have obtained the gymnasieskola degree,
and are either working or studying to obtain a Bachelor
Bachelor’s degree. Even though education level will degree:
Gymnasie
42,7%
not be tested for representativeness in the next skola:
paragraph, this variable could be used to test if there 29,1%
is a difference among the young adults in safety
perception with education level being the variable
tested for significance.

33
4.2.7 Public spaces and feeling unsafe
The first question of the survey discussed safety perception in a general way, asking the respondents
if they ever felt unsafe in public spaces. About half of the respondents tend to answer that they
occasionally feel unsafe when walking alone outside, and the other half does not feel unsafe at all
(table 4.1). However, this does not mean that half of the respondents cannot be used when analyzing
questions regarding public safety and feeling unsafe. This almost perfect division of the respondents
shows that there is not one universal opinion about safety perception.

Table 4.1 General safety perceptions respondents

4.3 Representativeness
In the previous paragraph several characteristics of the respondents were described, and compared
to the population of Stockholm. Descriptive statistics have the goal to give an image of the gathered
empirical information and compare this data to the existing statistical data, which in the case of this
study comes from the Swedish Statistical Database (2015). Three of the descriptive statistical
variables that were mentioned in the previous paragraph will be tested for representativeness. These
variables are: gender, age (divided into age groups) and ethnicity. These variables were chosen by
looking at what possible factors could influence the perception of safety and can be compared to the
existing statistical database. Factors such as occupation or educational level were not able to be
compared due to the different structure of the existing statistical database compared to the
gathered empirical data for this study (such as different educational categories, age groups etc.). The
three variables gender, age and ethnicity are also three variables that focus on the characteristics of
the respondents.

The testing for representativeness is done previous to the in depth analysis, because the
representativeness gives meaning to the in depth analyses; if the population is not representative,
any statistical statements made regarding the study will have little to no meaning.

In order to test the three selected variables for representativeness, a certain statistical test needs to
be selected, which is the Chi square (Goodness-of-fit) test. This test requires data to be transformed
(if necessary) into percentages, in order to compare percentages from the existing population with
the gathered date in this study. This means that absolute numbers cannot be compared, and have to
be transformed into groups and have percentages assigned to the groups. The outcomes of the
goodness-of-fit tests are summarized in table 4.2, and will be discussed in this paragraph.

34
Table 4.2 Representativeness of respondent characteristics

Source: SCB (2015) & Own research (2015)

4.3.1 Gender
The percentages of the respondent group and the population of Stockholm are relatively close to
each other, which could indicate that representativeness is quite likely. After applying the Goodness-
of-fit test the significance level outcome of 0,601 is higher than the set number of 0,05 which means
that there is no significant difference between the respondents and the population of Stockholm.
With 95% reliability it has been determined that the variable ‘gender’ is representative and can
therefore be used for making statistically generalizing statements.

4.3.2 Age
In the survey, respondents were asked to fill in the year in which they were born. In order to
calculate the age of the respondents in a quick and efficient manner, the year that the respondents
filled in was deducted from 2015. However, before the Goodness-of-fit test could be applied, the
different ages needed to be categorized into age groups. This is relatively easy and common to do
when studying people within a large age target group (for example 18-65 year olds), and becomes a
little more difficult when focusing on a smaller target age group, which is the case for this thesis. It
was decided that the target age group of 18-30 year olds was to be divided into three groups. The
first group consisted of 18 to 21 year olds, the second group 22 to 26 year olds and the last group 27
to 30 year olds. The groups do not consist of equal amount of ages, however as long as the
population of Stockholm group is also divided into identical groups this difference in ages among the
groups should not matter.

The age groups from the respondents differ somewhat from the age groups of the population of
Stockholm; but without testing little can be said regarding these differences. After applying the
Goodness-of-fit test it appears that with a significance level of 0,388 the variable age of the
respondents does not differ significantly from the population of Stockholm. This means that the age
variable can be used for making statistically generalizing statements as well.

4.3.3 Ethnicity
The population of Stockholm does not consist of solely people of Swedish decent. This means that in
order to have an overall representative sample population, this group needs to fit the characteristics
of the larger population. The variable ethnicity in this research may not be used for further in depth
analysis, but it is important to use as many variables as possible to determine the overall
representativeness of the sample population, which includes testing ethnicity for representativeness.
The amount of Swedes is higher among the respondents of the survey than the population of
Stockholm (figure 4.3). After testing the variable ethnicity with the Goodness-of-fit test, ethnicity
among the respondents does not significantly differ from the population of Stockholm, with 95%

35
reliability. The term ‘migrants’ used in table 4.2 refers to people that were not born in Sweden, and
have migrated to Sweden.

All three variables that were tested for representativeness do not differ from statistics of the
population of Stockholm from the SCB (2015). This means that further in-depth testing is possible,
and no weights need to be added to the variables.

36
5. In-depth analyses of the hypotheses and heat maps

In the previous chapter it was confirmed that the respondents to the survey were representative in
relation to the Stockholm population. This allows further statistical testing with other outcomes of
the survey. This chapter will cover the (possible) relations between spatial, social and individual
factors regarding safety perception from the hypotheses and conceptual model from chapter 2 and
3. The outcomes of these statistical tests will not only be mentioned but also discussed in the
perspective of the literature; e.g. what does it mean if two variables have a significant relation?

The previously made hypotheses will be discussed systematically using descriptive statistics to clarify
the relationship between the variables and the tests used. Using the results from the tests, a
hypothesis is either accepted or rejected, meaning that the relation between two variables is either
confirmed or denied.

This chapter consists of two parts: the statistical tests applied to the hypotheses and analysis using
the literature, and the answering of the sub questions in order to be able to answer the main
research question of this thesis in the concluding chapter.

The outcomes of the tests in this chapter can be found in appendix 8.4 of the thesis.

5.1 Hypothesis 1: Women perceive public spaces as less safe than men
This hypothesis is relatively straightforward and uses two variables: gender and feeling safe. These
two nominal variables can be tested using a Chi-square test, which will then determine whether or
not women feel less safe than men in public spaces. Earlier on in the theoretical chapter the
difference in safety perception among men and women was discussed. In general, men feel safer
than women in public spaces, meaning that it is to be expected that there should be a difference
among the respondents regarding this hypothesis. It appears that relatively speaking women feel
more frequently unsafe than men (figure 5.1). This means that among the respondents more women
than men have stated that they occasionally feel unsafe in public spaces. Among the men it appears
that there is quite an even balance between feeling safe and unsafe; a little more than half the male
respondents have stated that they do not feel unsafe at all.

Figure 5.1 The variables gender and feeling safe compared with each other

Do you ever feel unsafe when walking alone

outside?
100
90
80 42,86%
51,85%
70
60
50 Safe
40 Unsafe
30 57,14%
48,15%
20
10
0
Men Women 37
Without statistical testing it appears that there could be a difference between men and women
regarding the feeling of safety in public spaces. However, after applying the Chi-square test it
appears that these two variables are not statistically significant. The Chi-square test assumes that
there is no significant relation between two variables if the significance level of the test is higher than
the confidence level of 0,05. This means that in order to have a significant relation between the
variables, the outcome of the test needs to be lower than 0,05 (with a confidence interval of 95%).
For this test, the testing of gender in relation to feeling safe, the Pearson Chi-square value of 0,833
has a significance level of 0,361 (appendix figure 4.1).This is higher than the confidence level of 0,05,
which means that there is no significant relation between gender and feeling safe.

However, this does not mean the hypothesis that women perceive public spaces as less safe than
men is entirely rejected, as only one variable has been tested together with gender. Among other
type of questions, the survey contains a likert scale question (a question that consists of several
statements with which the respondents can either disagree, have no opinion of or agree with), of
which the statements together form a relatively good indication of whether or not the respondent
feels safe or unsafe in public spaces. The statements used to create one variable are: “most public
areas in Stockholm are unsafe”, “for my own safety, I would rather walk through Stockholm with a
friend/colleague than alone”, “I tend to avoid small/narrow streets when it is dark outside” and “I
tend to avoid public spaces when it is dark outside”. The data from the answers of these statements
have been computed into one variable. This variable, called “feeling safe in public places” was tested
with gender to find a possible significant relation. As this variable contains several questions that
question the behavior of the respondents, it should be noted that the outcomes of testing this
variable will say something about the (different) behavior between men and women, rather than
answering the hypothesis that women perceive public spaces as less safe than men.

Before using this newly created variable, it must be tested for internal consistency. This means that
the statements chosen must have similar outcomes. If people chose to totally agree with the first
statement but then totally disagree with the second statement, this means that internal consistency
is low. Using the Cronbach’s alpha test it is confirmed that there is a good internal consistency among
the four statements, meaning that they can be merged together into one variable (appendix figure
4.2.1).
The outcome of the Independent Samples T-test shows that there is a significant difference between
men and women regarding feeling safe in public spaces. The significance level of 0,035 is lower than
the confidence level of 0,05 (appendix figure 4.2). This means that using the likert scale test in order
to determine if there is a difference between men and women regarding feeling safe outside gives a
different result than the chi-square test.

The first hypothesis (“women perceive public spaces as less safe than men”) is partly rejected, which
conflicts with some of the literature found regarding this topic. Within the literature it was found
that men generally feel safer than women, but among the respondents of the survey in Stockholm
this is not the case. However, this does not mean that the variable gender is not suitable for further
investigation; for example women could appreciate different spatial factors than men regarding their
own safety. The second way to calculate this hypothesis resulted in a significant relation between
gender and behavior regarding safety perception. Using the likert scale questions, this hypothesis has
to be taken into account when answering the sub questions and the main question of this thesis.

5.2 Hypothesis 2: Environmental factors (such as the presence of trees, bushes, water etc.) in an
area are more important to women than men regarding safety perception

38
The second hypothesis looks at the difference between men and women and the values given to
environmental factors. Within the survey respondents were asked to rank which factors they value
the most regarding their own safety in public spaces. The selection that the respondents could
choose from was based on the literature discussed in the theoretic chapter of this thesis.
Figure 5.2 The outcomes of ranking the top three most important factors regarding own safety in public
spaces
Top three most important factors that affect
safety in public spaces
Lack of streetlights
5,88%
14,71%
Too much green in the area
17,32%
Areas that are difficult to get
9,48% an overview of
The presence of youths

14,05% The presence of beggars

19,61%
The time of day

18,95% Water present in the area

The respondents were able to choose from seven factors, which needed to be ranked from the most
important factor (ranked number 1) to the least important factor (ranked number 7). When looking
at the outcome of all respondents combined, thus taking men and women together, it appears that
environmental factors are not the most important (figure 5.2). It seems that the presence of the
‘unknown other’ is more important than environmental factors such as the presence of green in the
area, a lack of streetlights or even the time of day (Maruthaveeran & Konijnendijk van den Bosch,
2014).

Table 5.1 The factors ranked as number 1 regarding safety perception by men and women (absolute numbers)

39
The factors that the respondents marked as most important (ranked number 1) show that there is
quite some concern regarding the amount of green in public spaces (table 5.1). When looking at the
absolute numbers it shows that from both the men and women a significant amount of respondents
feel that too much green in an area has a negative effect on their safety perception. It must be noted
that no respondent chose to put ‘presence of water in the area’ as their number 1 choice, however
within the top three this factor has been chosen quite a lot (figure 5.2). After using the Chi-square
test to determine if there is a significant difference between men and women, and the factors
chosen to be most important regarding safety perception, it turned out that with a significance level
of 0,056 there is no significant difference between men and women (appendix figure 4.3). The
options chosen by men and women are too similar. Even though the outcome of the test shows that
there is no difference between men and women, the descriptive statistics have shown the
importance of the presence of green in public spaces, and how this affects safety perception.

The second hypothesis is rejected, as there is no difference between men and women and how they
value environmental factors in their safety perception, but it could be stated that in general men and
women both feel that too much green in an area is associated with possible danger and feeling
unsafe.

5.3 Hypothesis 3: Women experience night time as more threatening to their safety than day time,
compared to men
From the literature it became clear that night time, or lack of daylight, has a direct influence on
people’s safety perception (Brands et al., 2015; Maruthaveeran & Konijnendijk van den Bosch, 2014).
People’s ability to get an overview of their surroundings is limited and the association between night
time and crime is often made, which could make some people feel uneasy at night (England & Simon,
2010). Among the respondents to the survey of this study, it appears that there is a difference to be
found between men and women (figure 5.3). Men tend to not feel bothered by the darkness
regarding their own safety perception, with 74% stating that it does not matter if it is day time or
night time when talking about their own safety. When comparing the outcomes of the women to the
men it appears that women tend to agree more with the posed statement “I tend to avoid public
spaces when it is dark outside”.

Figure 5.3 Comparing men and women regarding the avoidance of public spaces when it’s dark outside

Would you rather avoid public spaces when it's

dark outside?
100,00%
90,00%
80,00%
70,00% 74,07%
60,00%
50,00% Agree
53,06%
40,00% 46,94% Disagree
30,00%
20,00% 25,93%
10,00%
0,00%
Men Women 40
After using the Chi-square test it appears that there is indeed a significant difference between men
and women and whether or not public spaces would be avoided during night time. The Pearson Chi-
square score of 4,928 with a significance level of 0,026, shows that with 95% reliability this
hypothesis is accepted (appendix figure 4.4). Women experience night time as more threatening to
their safety than day time, compared to men. When taking this statement into a larger perspective,
and into the research area of this study, Stockholm, this could have a major influence on the mental
maps people create of the area they visit on a regular basis (England & Simon, 2010, pp. 202-203).
People create their own mental maps about places they know of and visit, and several external
factors influence these mental maps. However, the concept of a mental map is not static. Darkness
or high crime rates can influence these mental maps and can create zones in which people would go
as long as there’s daylight, but avoid when night time sets in. The outcome of this hypothesis is that
women would rather avoid public spaces when it is dark outside, which could point towards a
negative association to the winter season by women, regarding their own safety.
The strength of the relation between the tested variables can be calculated using the Cramér’s V test.
The outcome of the test shows that with a score of 0,219 there is a weak to relatively strong relation
between gender and avoiding public space when it is dark (appendix figure 4.4.1). When looking at
gender there is no point to calculate the direction of the relation of the variables, you are either a
man or a woman. This is a variable that cannot change over time, and therefore the direction of the
opinion would be irrelevant to calculate using the Kendall’s tau test. The Kendall’s tau test would be
relevant when comparing the amount of time people lived in or near Stockholm to safety perception;
time could influence the way people perceive their own safety in public spaces.

5.4 Hypothesis 4: Different seasons have a direct influence on the safety perception for both men
and women
This hypothesis takes the outcomes of the previous hypothesis and puts it in a larger perspective; a
difference between men and women has been confirmed regarding the role of daylight and safety
perception, but do different seasons have a similar impact?
There are several ways of determining what influences different seasons have on the safety
perceptions of people. Within the survey a likert scale style question was posed to the respondents
regarding the use of public spaces during different seasons. The outcomes from this question show
that both men and women tend to make more use of public spaces during the summer than in the
winter. This is not very surprising, however after testing this variable between men and women it
does confirm that there is no significant difference between men and women regarding their opinion
of use of public spaces in different seasons (figure 5.4).
Figure 5.4 Use of public space during summer and winter between men and women

41
Even though there is no difference to be found between men and women in the use of public spaces,
the descriptive statistics show the importance of seasons. The summer is apparently much more
inviting towards people to make use of parks, squares and other public spaces. About 15% of the
men and women state that it does not matter if it is winter or summer regarding their use of public
spaces.

In order to look further into the behavior of people in different seasons, and their perception of
safety, it is needed to combine variables together. The abovementioned variable that looks into if use
of public spaces is affected by seasons can be merged with another variable. In the same part of the
survey, respondents are asked their opinion on the statement “I feel safer in public spaces during the
summer than during the winter”. If these two variables are put together the assumption is made that
feeling safe in the summer and making more use of public spaces in the summer occur together and
can be used as one variable. In order to test the internal consistency for two variables the Spearman-
Brown test is more suitable than the previously used Cronbach’s alpha test (de Vocht, 2010). The test
showed that there was enough internal consistency to merge these variables into one variable for
further testing (appendix figure 4.5).

Figure 5.5 The seasonal influence on safety perception and use of public spaces

Do you feel safer during the summer in public

spaces, and does this influence your use of public
spaces?
100,00%
90,00%
80,00%
70,00% 77,55%
60,00%
50,00% 59,26% Agree
40,00% Disagree
30,00% 40,74%
20,00%
22,45%
10,00%
0,00%
Men Women

When this variable consisting of two similar variables is tested, a difference between men and
women can be seen (figure 5.5). Among the women it appears that the vast majority feels safer
during the summer, and that this feeling of safety affects their use of public spaces in a positive way.
Women make more use of public spaces during the summer than in the winter, because of an
increased feeling of safety during the summer. Summer is often associated with longer availability of
daylight and warm, sunny weather. Winter is often seen as cold, dark and not very inviting towards
public spaces. Men also feel relatively positive towards the summer and how the summer increases
their use of public spaces, but not as much as women do.

After using the Chi-square test the outcomes show that there is a significant difference between men
and women and the way safety and use of public space during the summer are perceived, with 95%
reliability. The Pearson Chi-square score of 3,948 with a significance level of 0,047 confirms the

42
difference between the two groups (appendix figure 4.6). In order to determine the strength of the
outcome of the Chi-square test, a Cramér’s V test needs to be performed. The Cramér’s V test shows
that there is a weak relation between the tested variables (appendix figure 4.6.1). This means that
the difference between men and women regarding their safety perception and use of public spaces
during the summer is significant, but not very strong. According to the literature from chapter two in
this thesis the difference between men and women regarding safety perception had been confirmed,
but not in the light of how seasons could potentially influence their behavior. To calculate the
direction of how the relation between the variables develops itself is once again not relevant due to
the testing of gender. This means that the Kendall’s tau test is irrelevant.

Concluding, the outcomes of these tests mean that the hypothesis “Different seasons have a direct
influence on the safety perception for both men and women” can be accepted, with the note that
there is a significant difference between men and women. The difference between men and women
that has been tested confirms the outcome of the study done by Brands et al. (2015). This study
asked respondents (with a similar target group to this study, consisting of young adults) to describe
the way they felt about certain areas illustrated with photographs, mostly at night time. There was a
difference between men and women and how they perceived their own safety in the different
situations; men tended to be more positive and less anxious about the photographs, while women
felt that they would be unsafe in the selected pictures because of the possibility to become a victim
of crime.

5.5 Heat map results

Two questions of the survey were posed as situational (hypothetical) questions, asking the
respondents to mark certain areas within the selected research area that they would rather avoid
during a certain time of the day and during a specific season. This information could pinpoint places
within the research area that the respondents find uncomfortable to be in. Additionally, the
respondents were asked why they felt uncomfortable in the area(s) they selected. As the heat map
questions were not obligatory for the respondents to fill in, not every respondent has marked a spot
on the map that they feel unsafe in. Due to the literature pointing towards a difference in safety
perception between winter and summer time, a difference in response between the two seasons is
expected. The amount of people that responded to the winter map was 68, and the amount of
people that responded on the summer map was 42. Each respondent had a maximum of three spots
they could mark, however not every respondent made use of this maximum.

When comparing the heat maps of the two seasons to each other, it appears that the winter map is
more scattered than the summer map. The summer map has a strong concentration in the area near
Kulturhuset. Additionally, the area near Kungsträdgården is also selected often in the summer. This
means that these two areas in particular, during the summer time, are experienced as unsafe at
20.00 o’ clock. The respondents feel that there are more areas within the selected research area that
are unsafe than in the summer; this statement does not necessarily mean that the winter season is
more unsafe than the summer, as it was determined that hypothesis 4 (‘Different seasons have a
direct influence on the safety perception for both men and women’) was rejected. It appears that the
area closest to the central station of Stockholm is often perceived as unsafe by respondents in the
winter. In the summer some of the areas near central station are still perceived as unsafe at 20.00 o’
clock, but less so than in the winter.

43
Figure 5.6 Winter heat map: selected areas respondents would avoid at Figure 5.7 Summer heat map: selected areas respondents would avoid
20.00 at 20.00

The selected heat maps show only a part of the research area; this is done because other parts in the
research area were not selected by many people, thus less interesting when analyzing this spatial
data (figure 5.6 & 5.7).

The heat maps and the corresponding reasons why people chose the areas on the map are not used
for in-depth statistical testing. Instead, the reasons that people chose to avoid the areas during the
different situations will be looked at and compared. The descriptive statistics show that there is a
shift in why people tend to avoid certain areas. The reason why 20.00 o’clock was chosen for both
scenarios (winter and summer) is because of the change in daylight between the two seasons. During
the winter there is less daylight than in the summer, which resulted in the assumption that there
would be differences in the areas people would avoid, and the reasons why they would avoid these
areas. During the winter, the respondents have answered that the lack of streetlights and lack of
visible security in areas are the main factors in determining their level of safety. However there were
a lot of respondents that had a specific reason that was not listed among the multiple choice
answers. One of the most given answers was that there were not enough people around that could
possibly help in case someone were to be a victim of a crime. This relates to the literature regarding
‘safety by community’, about how keeping an eye on each can decrease crime rates in areas
(Schroeder & Anderson, 1984; Mehta, 2014). It appears that among the other answers given for the
winter season, the respondents feel uneasy about other people, especially beggars or drunks. This
‘fear of the unknown other’ is a concept discussed earlier in the theoretic chapter as well, which is
not necessarily a spatial related factor, but more of a social factor (Maruthaveeran & Konijnendijk
van den Bosch, 2014). Comparing the winter descriptive statistics to that of the summer, it appears

44
that there are some differences; for example lack of streetlights only has 3,8% of the total amount of
responses, while lack of visible security in the area has become bigger. This means that the lack of
daylight is not a very much of a factor anymore; however the change in season invites more people
to make use of public spaces. According to the respondents, this process requires more visible
security in order to make them feel safer in public spaces. The answers given by the respondents that
they could voice with the ‘other’ option shifted to being scared of becoming a victim of getting
harassed by certain groups of people. This is a difference when compared to the winter season:
during the winter the respondents feel that there are not enough people around to keep an eye on
each other, and during the summer people feel that there are too many people that could potentially
form a threat.

Figure 5.8 Summer and winter hypothetical situation: reasons why people avoid certain areas

Reasons avoiding area during the winter

at 20.00 o' clock
Lack of streetlights

Lack of visible security in

18,11% the area
22,05%
Too much security in the
area
Narrow streets

15,75% 20,47% Too many trees or bushes

in the area
Badly maintained area
8,66%
12,60% Other reason
2,36%

Reasons avoiding area during the summer

at 20.00 o'clock Lack of streetlights
3,80%

Lack of visible security in

16,46% the area
Too much security in the
25,32% area
Narrow streets
20,25%
Too many trees or
bushes in the area
1,27%
Badly maintained area
21,52%
11,39%
Other reason

45
5.6 Extra testing
This paragraph contains tests and analyses that do not belong to a specific hypothesis but could be of
value when concluding the sub-questions and main research question.

5.6.1 Safety perception in different seasons among different residents Stockholm

The variables that are tested here tell if the amount of time lived in or near Stockholm affects an
individual’s safety perception. For example if someone would live longer in Stockholm, it is to be
expected that this person would be more familiar with the city and thus feel safer in public spaces
than someone who has not lived as long in the city.
At the end of the survey respondents were asked to fill in how long they’ve lived in or near
Stockholm, which is one of the variables used here. The other variable is a likert-scale question which
has been used for some of the previous hypotheses: “I feel safer in public spaces during the summer
than during the winter”. For the testing of these variables it must be noted that no difference
between men and women has been made. Instead the focus lies upon the possible difference
between the amount of time lived in or near Stockholm in relation to seasonal safety perception.
Figure 5.6 The influence of time lived in Stockholm on safety perception during different seasons

The residents living longer in or near Stockholm feel differently towards the two seasons and how
this affects their safety perception. The majority of the people living for ten years or longer in
Stockholm think that feeling safe does not change whether it is summer or winter. The people living
in Stockholm from one to five years have the opposite opinion; summer does affect the safety
perception and makes them feel safer. These two variables have been tested using the Independent
Samples T-test, which showed that there is a significant difference between the two inhabitant
groups with 95% reliability (appendix figure 4.7). When taking the target group for the survey (18 to
30 year olds) into account, the inhabitants that have lived longer than ten years in Stockholm have
most likely grown up in the city. The inhabitants that have lived between one to five years in
Stockholm have moved relatively recently in their lives, and it could be argued that this has been
done for work or study related reasons.

Concluding it could be said that the longer one lives in Stockholm, the more familiar one becomes
with the city, which influences the safety perception throughout different seasons in a positive way.

46
5.6.2 Visiting frequency to the city in relation to avoiding narrow streets when it is dark outside
The data from the survey regarding the frequency with which the respondents visit the city of
Stockholm could offer insight into the relation between safety perception and familiarity with the
city. Similar to the test with the variable that looked at the amount of time respondents had lived in
Stockholm, the variable visiting frequency looks at commuters and inhabitants together. The
respondents were asked to answer how frequently they visit the city of Stockholm, in a multiple
choice setting. The possible answers consisted of: 1 time per month, 2 times per month, 1-2 times a
week, 2-4 times a week, 4-6 times a week, or daily. In order to be able to test this variable a division
of the answers was made into ‘infrequent’, ‘regular’ and ‘frequent’ visitors. The people that visit the
city 1 or 2 times per month are categorized as infrequent visitors. The people that visited the city 1 to
4 times a week are categorized as regular visitors. The last group, labeled as frequent visitors,
consists of the people that visited the city 4-6 times a week to daily.

When comparing the visitor groups in relation to how they feel about avoiding narrow streets or
small spaces when it is dark outside shows a significant difference between the groups (figure 5.7).
The Chi-square test shows, with a Pearson Chi-square value of 12,486 and a significance level of
0,002, that there is a significant difference between the amount of visits and the way narrow streets
and small spaces are perceived when it is dark outside (appendix figure 4.8). The Cramér’s V test
shows that there is a relatively strong relation between the two variables; however this does not say
anything about the direction of this relation. The Kendall’s tau-c test shows that the relation between
the two variables develops in the following way: the more a person visits the city, the less impact
small or narrow streets have on the safety perception of the individual when it is dark outside
(appendix figure 4.8.1).

Figure 5.7 The frequency of visits by respondents in relation to the answer given to the statement ‘I would
avoid narrow streets when it’s dark outside’

Visiting frequency Stockholm city in relation to

the statement 'I would avoid narrow streets
when it's dark outside'
50%
45%
39,22%
40%
35%
30%
25% Agree
18,63%
20% Disagree
15% 13,73%
10,78% 9,80%
10% 7,84%
5%
0%
Infrequent Regularly Frequent

47
6. Conclusion

Feeling safe in public spaces is a relatively well known subject among urban researchers. It is a
subject that often has its focus on the entire population. From these existing studies, there appears
to be a big difference between young and older people regarding safety perception. Younger people
tend to feel safer than older people, and men tend to feel safer than women. By gathering empirical
data regarding the safety perception of younger people, this thesis tried to get a better
understanding of this group of relatively young people and how they perceive their own safety. Using
a survey that was spread both online and in the field itself, respondents were able to voice their
opinion on hypothetical situations and factors that they believe are vital for their safety. By selecting
a research area in the city centre of Stockholm, location specific questions could be asked to the
respondents, which in its turn could help pinpoint areas that need extra attention in improving the
feeling of safety. The data gathered was transformed into a statistical database in order to test
variables that could point towards answers regarding young adults and the way they perceive their
own safety. The statistical tests would determine if the hypotheses that were developed in the
methodology chapter, based on the literature, were to be accepted or rejected. These hypotheses, in
their turn, would help answer the sub-questions, which were made in order to be able to answer the
main research question of this study: “In what ways does safety perception affect the use of public
spaces by young adults?”

The first sub-question is “in what ways does safety perception differ between men and women?”,
which can be answered by looking at the outcomes of several of the hypotheses. After testing it
appears that there is no significant difference between men and women in relation to feeling unsafe
outside. Even though women tend to feel a little more unsafe then men, after testing it became clear
that this difference was not significant and thus cannot be accepted. However, when the behavior of
the two groups was tested, a significant difference was found. Using several statements regarding
behavior that the respondents could voice their opinion on, a distinction between people that felt
safe and people that did not feel safe was made. Statements such as “for my own safety I would
rather walk through Stockholm with a friend/colleague than alone” and “I tend to avoid
small/narrow streets when it is dark outside” were mostly agreed upon by women, while men
tended to either disagree or feel neutral about these statements. Regarding the behavior of men and
women there was a significant difference, which means that men and women see situations
differently in regards to how it can affect their own safety. This significant difference could explain
why the first hypothesis was not accepted; if men and women have their own concept of ‘feeling
unsafe’, then a general statement is perhaps too vague in order to determine a difference between
men and women. According to the outcome of the behavior test, women tend to deal with situations
differently than men, perhaps avoiding certain possible situations or areas. This process of avoiding
possible situations and areas that could compromise the safety could help women into feeling safe in
the areas they do know. When looking at factors that could influence the safety of an individual,
based on the literature of this study, there appears to be no difference between men and women.
The respondents could value several factors that could affect their safety by ranking them from most
important to least important. Most men and women felt that ‘too much green in the area’ was the
most important factor that affects their safety perception. Areas that are difficult to get an overview
of, was the second most picked factor for women. This means that if an area consists of small streets
or a lot of green, this affects the ability for people to determine whether or not an area is safe or if an
area is unsafe. Another difference between the two groups is that men tend to not avoid public
spaces when it is dark outside, as opposed to women. After testing it appeared that there is a
significant difference between men and women in relation to darkness and the use of public space.
This darkness is something that affects a city like Stockholm more in some seasons than in others.

48
This difference in daylight between the winter and the summer season was looked at within the
survey as well. The tests showed that men and women make more use of public spaces during the
summer. However when looking at the combined statement: ‘do you feel safer during the summer in
public spaces, and does this influence your use of public spaces?’, a significant difference was
determined between men and women. Women agreed significantly more with this statement than
men, meaning that the summer does not affect men’s safety and use of public spaces as much as it
affects women’s perception.

The second sub-question is “which factors seem to be the most important with reference to safety
perception in general (by both men and women)?”. As had been determined in the previous sub-
question, different seasons have an effect on the safety perception of women. Too much green in
areas can compromise the ability for people to estimate if an area is safe or unsafe, and thus are
these areas often experienced as unsafe. In the previous tests men and women were taken as two
separate groups, however extra tests were made in order to test different groups of people. Within
the survey people were asked to fill in how long they’ve been living in or near Stockholm. This
variable was tested with feeling safe in different seasons. Residents that had lived in or near
Stockholm for longer than ten years appeared to feel no different towards the winter season than
the summer season. The residents that had lived in Stockholm between one to five years felt
significantly safer in the summer season than in the winter season. This indicates that familiarity with
the area has a positive influence on the safety perception of Stockholm residents, eliminating the
preference for a certain season regarding their own safety. Within the survey respondents were also
asked how often they visit the city. This variable was tested with the statement “I would avoid
narrow streets when it is dark outside”, in order to see if there was a difference between the
frequency of visits to the city and feeling unsafe in particular areas of the city. After testing it
appeared that the frequent visitors had a significantly different opinion than the infrequent of
regular visitors. Frequent visitors felt that there was no need to avoid narrow streets when it is dark
outside, unlike the infrequent and regular visitors. After testing these two variables on how they
develop, it appears that the more often an individual visits the city the less impact small areas have
when it is dark outside.

The third and last sub-question is “how can public spaces be improved in order to grow a positive
sense of safety among the visitors?”. The results from the heat maps gathered with the survey it
appears that among the respondents the area near Kulturhuset is often associated with a lack of
visible security in the area and a badly maintained area. According to the respondents this is also the
area they would rather avoid during the winter season when it is dark. However, some other areas
were marked by the respondents as well as unsafe, mostly having to do with the lack of other people
around and the lack of streetlights which compromises the visibility of an area. Small street areas in
or near Gamla Stan, the small peninsula Skeppsholmen and some parts of Djurgården are examples
of this. When comparing the winter to the summer map, the areas that were often marked during
the winter were less visible on the summer map. This meant that there was a change in areas the
respondents would avoid. The area near Kulturhuset and the central station contained less scattered
marked areas in the summer map as places the respondents would avoid than on the winter map.
Additionally, respondents were able to write down their own specific answer if their answer was
different than the multiple choice options given to them. A substantial amount of respondents felt
that they would avoid areas (both during the summer and winter) because of drunks and beggars,
which had an impact on their safety. This relates back to the literature regarding ‘fear of the
unknown other’.

49
To answer the main research question “In what ways does safety perception affect the use of public
spaces by young adults?”, one must look at the outcome of the different sub-questions. Men and
women experience safety as different processes in general, not just when looking at the entire
population, but also when looking at young adults. Men tend to feel more indifferent towards certain
situations than women. This does not necessarily mean that men are safer than women, according to
the statistical research discussed in the theoretic chapter; men are more often victims of crime than
women. However the differences in potential crimes also affect women, as women are more often
subject to sexually related crimes than men, which has an impact on their safety perception. In
general young adults make more use of public space during the summer and tend to avoid areas that
contain too much green and are difficult to get an overview of. Additionally, for young adults living in
Stockholm for ten years or longer the seasons have less of an impact than for young adults living in
Stockholm for a shorter period of time. Darkness and seasons are two factors that have an influence
on the safety of certain groups within the young adults. Women tend to feel safer during the
summer, which affects their use of public spaces in a positive way. Infrequent and regular visitors to
the city of Stockholm are more intimidated by darkness than frequent visitors. Some areas within the
selected research area are more popular during the summer than during the winter, which is
affected by the amount of people in a certain area and the amount of daylight available. Regarding
safety perception the young adults group is not a homogeneous group. Men and women behave
differently regarding their safety perception; women feel safer with the presence of a friend or
colleague when walking around Stockholm, while the presence of a friend or colleague does not
affect the safety perception for men. Young adults that have lived for a longer time in Stockholm are
not affected much by the different seasons, while young adults that have not lived as long in
Stockholm still prefer summertime over wintertime regarding their own safety.

The research question asked which factors affect the safety perception by young adults in general.
The results from this study show that within the group of young adults there are many different
groups that can be distinguished that each have their own opinion on safety perception that differs
significantly from other groups. The most obvious sub-group is perhaps men and women, but other
sub-groups such as visitors to the city, inhabitants of the city and Swedes and migrants (born outside
of Sweden) perceive their safety in significantly different ways. From existing literature the main
conclusion is that young adults feel significantly safer in public spaces than older people. This study
has confirmed that within the group of young adults, there is not homogeneity regarding safety in
public spaces, and instead spatial and social variables affect the sub-groups within the young adult
population of Stockholm in different ways.

50
7. Discussion

This chapter will reflect back upon the processes that came along with creating this thesis. First, the
methods used will be discussed, subsequently the use of the conceptual model will be discussed and
lastly the results from this study will be put in a broader perspective using the literature from the
theoretic chapter.

To be able to make generalizing statements about populations, quantitative studies are necessary.
The main goal of this study was to be able to see how the young adult population sees their own
safety perception in public spaces, and if there are any differences within this population. The chosen
method of conducting surveys offered a systematic way to gain information about a relatively large
portion of the target population, by spreading the survey online and by in-field surveying. This
method of surveying was accompanied by statistical analysis of the gathered data, which in its turn
enabled statistical testing and being able to make generalizing statements. Even though this method
offered a good way to gain insight in the target population, quantitative methods also have their
downside. Qualitative methods have their focus on a smaller scale such as neighborhoods or
individuals, which often offer more detailed insight into the chosen populations. In the case of this
study, focus groups could have offered insight into specific neighborhoods in the research area,
offering the participants a platform where safety perception could be discussed. If this study were to
be continued or conducted in a different city, a mix of quantitative and qualitative methods would be
preferred, due to the valuable insights of both methods.
When using statistical analyses, a common method is to create hypotheses that are derived
from existing literature. These hypotheses are assumptions from the researcher (based on existing
literature and/or assumptions) which suggest certain relations between variables in a study. For this
study, hypotheses regarding gender and safety perception were developed in order to answer the
sub-questions. Additional testing between other variables was done in order to gain more
information about young adults as a group, instead of only looking at the differences between men
and women. The survey consisted of several likert-scale questions (questions with statements where
the respondents can agree or disagree with), which were very valuable when making statistical tests
in order to accept or reject the majority of the hypotheses. If this same study were to be done again,
the survey could be adjusted slightly in a way that it would contain more likert-scale questions, as the
outcomes from these questions offer a direct insight into the opinion of respondents on hypothetical
situations. When conducting a survey, the option for open answer style questions must be added,
however the downside of open answer questions in a quantitative survey is that these answers
cannot be statistically tested. The survey that was used for this research was often well received by
the respondents; there were only a few people that commented that it was too long for them to
finish the survey.

The survey questions and the hypotheses were based upon existing literature. The existing literature
pointed towards vital relations between several factors that had an impact on safety perception.
These variables and relations were visualized into a conceptual model (figure 7.1). The main goal of
this study was to look further into the relations between the variables discussed in the theoretic
framework chapter and the conceptual model (for example how men and women (gender) would
relate to different seasons, or the green infrastructure in Stockholm regarding their safety
perception). The conceptual model helped with keeping an oversight on which factors and variables
were of importance and which tests needed to be done in order to get a complete insight into the
young adult population regarding the topic of safety perception in public spaces. As stated earlier,
the variables that could possibly affect the safety perception of people were derived from existing
literature. However, if interviews or focus groups were held prior to the creation of the survey,

51
perhaps factors that would come up during these interviews or focus group meetings could have
been added to the list in the conceptual model.

Figure 7.1 Conceptual model derived from existing literature made by the author (2015)

The statistical testing showed that within the group of young adults many sub-groups could be
distinguished, for example: men and women, visitors to the city, inhabitants of the city and Swedes
and migrants. The statistical outcomes showed that there are not several set factors that have the
same influence on all of the sub-groups. For example, a significant difference between men and
women could not be seen when testing ‘gender’ with the variable ‘avoiding narrow streets when it is
dark outside’. However, when this same variable was tested with the sub-group ‘visitors’ (infrequent-
, regular- and frequent visitors) a significant difference within this group could be seen. These
outcomes show that safety perception is a complicated topic of which the spatial and social factors
have different effects on different groups of people. When relating the results back to the literature
in the theoretic framework chapter, it appears that only few studies actually looked at the behavioral
differences between men and women. Shenassa et al. (2006) asked respondents if they would feel
safe to walk around their neighborhood when it is dark outside, which is a hypothetical behavioral
question. The behavior of the respondents to the survey in this study was also tested, and similar
results were found, compared to the study of Shenassa et al. (2006). Men and women behave
differently when being in public spaces, and feel differently about their own safety. This does not just
confirm the study done by Shenassa et al., but also the other literature which does not necessarily
test on behavior differences between men and women. This similar behavior among the entire
population, and not just among young adults, shows that behavioral patterns are not limited or
changed depending on age. Within the literature, little was discussed regarding safety perception in
relation to different seasons. This is why the outcomes of studies that looked at the influence of
daylight on safety perception were used in order to determine if seasons could have the same

52
influence within this study. The respondents to this study showed that there are different social and
spatial reasons for the summer and winter to avoid certain public spaces. Women tend to feel
significantly safer during the summer which influences the use of public spaces in a positive way,
while seasons appear to have less of an influence on the safety perception by men. Additionally, the
summer and winter seasons affect inhabitants of the city in different ways as well. People that have
lived for a shorter time in the city feel less safe during the winter than in the summer. The
inhabitants that have lived for a longer period of time in Stockholm are less bothered by the different
seasons regarding their safety perception. This relation between darkness and the amount of time
lived in the city could make the conclusion that the more familiar an individual gets with an area (or a
city in this case), the more positive one’s safety perception becomes. The measurement of familiarity
of a city linked to safety perception is relatively hard to find among the existing literature, however,
according to the outcomes of this study it could be argued that this relation of variables could be
taken into account when conducting similar studies.

The outcomes of this study show that safety perceptions in public spaces differ among the sub-
groups within the young adult target group. The lack of streetlights in certain areas of Stockholm city
during the winter, and the crowdedness and lack of visible security during the summer play
significant roles for young adults regarding their safety perception and avoidance of certain public
spaces. If this study were to be continued and taken into a broader perspective, a comparison could
be made between young adults and the rest of the population, to test the results from existing
literature regarding the safety perception of young and older people. Additionally, the same study
could be done in other cities (within Sweden or Scandinavia) in order to determine if young adults
feel the same way as they do in Stockholm, or if there are differences regarding safety perception
among different young adult populations.

53
References

Arendt, H. (1958), The Human Condition, Chicago, IL. University of Chicago Press

Bennett, G.G., L.H. McNeill, K.Y. Wolin, D.T. Duncan, E. Puleo & K.M. Emmons (2007), Safe To Walk?
Neighborhood Safety and Physical Activity Among Public Housing Residents, PLoS Medicine, Vol. 4, No. 10, pp.
1599-1607

Brands J., T. Schwanen & I. van Aalst (2015), Fear of crime and affective ambiguities in the night-time economy,
Urban Studies, Vol. 52, No. 3, pp. 439-455

Bromley, R.D.F. & R.J. Stacey (2012), Feeling unsafe in urban areas: exploring older children’s geographies of
fear, Environment and Planning A, vol. 44, pp. 428-444

Brå (2014), The Swedish Crime Survey 2014: Concerning exposure to crime, insecurity and confidence, English
summary of Brå report 2015:1

Burgess, J., C.M. Harrison & M. Limb (1988), People, Parks and the Urban Green: A Study of Popular Meaning
and Values for Open Spaces in the City, Urban Studies, Vol. 25, pp. 455-473

Chicago manual of Style (2010), The Chicago Manual of Style Online, The University of Chicago

England, M.R. & S. Simon (2010), Scary cities: urban geographies of fear, difference and belonging, Social &
Cultural Geography, Vol. 11, No. 3, pp. 201-207

Ferraro, K.F. & R.L. LaGrange (1992), Are Older People Most Afraid of Crime? Reconsidering Age Differences in
Fear of Victimization, Journal of Gerontology: Social Sciences, Vol. 47, No. 5, pp. 233-244

Gau, J.M., N. Corsaro & R.K. Brunson (2014), Revisiting broken windows theory: A test of the mediation impact
of social mechanisms on the disorder-fear relationship, Journal of Criminal Justice, Vol. 42, pp. 579-588

Google Maps (2015), Selected research area [online], http://www.google.se/maps

‘t Hart, H., H. Boeije & J. Hox (2009), Onderzoeksmethoden, Boom uitgevers Amsterdam

Kungsträdgården (2015), Kungsträdgården: Parken då och nu [online],

http://www.kungstradgarden.se/web/Parken_da_och_nu.aspx

Kuo, F.E., M. Bacaicoa & W.C. Sullivan (1998), Transforming Inner-city Landscapes: Trees, Sense of Safety, and
Preference, Environment and Behavior, Vol. 30, No. 1, pp. 28-59

Liempt, van, I., I. van Aalst & T. Schwanen (2015), Introduction: Geographies of the urban night, Urban Studies,
Vol. 52, No. 3, pp. 407-421

Lindgren, T. & M.R. Nilsen (2011), Safety in Residential Areas, Journal of Economic and Social Geography, Vol.
103, No. 2, pp. 196-208

Maruthaveeran, S. & C.C. Konijnendijk van den Bosch (2014), A socio-ecological exploration of fear of crime in
urban green spaces – A systematic review, Urban Forestry & Urban Greening, Vol. 13, pp. 1-18

Mehta, V. (2014), Evaluating Public Space, Journal of Urban Design, Vol. 19, No. 1, pp. 53-88

Pain, R. (2000), Place, social relations and the fear of crime: a review, Progress in Human Geography, Vol. 24,
No. 3, pp. 365-387

54
Perkins, D.D., J.W. Meeks & R.B. Taylor (1992), The physical environment of street blocks and resident
perceptions of crime and disorder: implications for theory and measurement, Journal of Environmental
Psychology, Vol. 12, pp. 21-34

Ruijsbroek, A., M. Droomers, P.P. Groenewegen, W. Hardyns & K. Stronks (2015), Social safety, self-rated
general health and physical activity: Changes in area crime, area safety feelings and the role of social cohesion,
Health & Place, Vol. 31, pp. 39-45
1
SCB (2015), Population by region, marital status, age and sex (Selections: region – Upplands Väsby, Vallentuna,
Järfälla, Ekerö, Huddinge, Tyresö, Täby, Danderyd, Sollentuna, Stockholm, Södertälje, Nacka, Sundbyberg,
Solna, Lidingö & Sigtuna, age – 18-30 years, year – 2014) [online],
http://www.statistikdatabasen.scb.se/pxweb/en/ssd/START__BE__BE0101__BE0101A/BefolkningNy/?rxid=945
91693-c24d-4973-8db1-426f99bf3d82
2
SCB (2015), Number of persons by region, age, year and foreign/Swedish background [online],
http://www.statistikdatabasen.scb.se/pxweb/en/ssd/START__BE__BE0101__BE0101Q/UtlSvBakgTot/?rxid=9ff
63151-2d55-4df6-a432-7c6ba2f5d857
3
SCB (2015), Average age of the population by region and sex [online],
http://www.statistikdatabasen.scb.se/pxweb/en/ssd/START__BE__BE0101__BE0101B/BefolkningMedelAlder/
?rxid=11ad7ca4-d1b8-4272-ae9c-ea793690e9ce
4
SCB (2015), Population by region, age, year and sex [online],
http://www.statistikdatabasen.scb.se/pxweb/en/ssd/START__BE__BE0101__BE0101A/BefolkningNy/?rxid=84e
05169-6f25-4f00-a149-06d4b60109bb

Schroeder, H.W. & L.M. Anderson (1984), Perception of Personal Safety in Urban Recreation Sites, Journal of
Leisure Research, Vol. 16, No. 2, pp. 178-194

Shenassa, E.D., A. Liebhaber & A. Ezeamama (2006), Perceived Safety of Area of Residence and Exercise: A Pan-
European Study, American Journal of Epidemiology, Vol. 163, No. 11, pp. 1012-1017

Stafford, M., T. Chandola & M. Marmot (2007), Association Between Fear of Crime and Mental Health and
Physical Functioning, American Journal of Public Health, Vol. 97, No. 11, pp. 2076-2081

The Guardian (2014), Anti-homeless spikes are just the latest in ‘defensive urban architecture’ *online+,
http://www.theguardian.com/cities/2014/jun/12/anti-homeless-spikes-latest-defensive-urban-architecture
[Quoted on 26-02-2015]

Thomas, M. (1991), The Demise of Public Space, Town Planning Responses to City Change, Avebury: Aldershot,
pp. 209-224

Valentine, G. (1990), Women’s Fear and the Design of Public Space, Built Environment, Vol. 16, No. 4, pp. 288-
303

Visit Sweden (2015), Time & Daylight Hours [online], http://www.visitsweden.com/sweden/Sweden-

Facts/Worth-knowing-about-Sweden/Time--daylight-hours/ [Quoted on 10-03-2015]

Vocht, de A. (2010), Syllabus Statistiek: Sociale Geografie en Planologie, Faculteit Geowetenschappen,

Universiteit Utrecht

55
Appendix

8.1 Research area

56
8.2 Survey

57
58
59
60
61
62
63
64
65
66
8.3 Representativeness tests (chapter 4)

Figure 3.1 Chi-square test for testing the representativeness of the variable gender in relation to the target population.

Test Statistics

What is your
gender?
Chi-Square a
,274
df 1
Asymp.
Sig. ,601

a. 0 cells (0,0%) have

expected frequencies less
than 5. The minimum
expected cell frequency is
51,3.

Figure 3.2 Chi-square test for testing the representativeness of the variable age in relation to the target population.

Test Statistics

Age into
classes
Chi-Square a
1,891
df 2
Asymp.
Sig. ,388

a. 0 cells (0,0%) have

expected frequencies
less than 5. The
minimum expected cell
frequency is 19,9.

Figure 3.3 Chi-square test for testing the representativeness of the variable ethnicity in relation to the target population.

Test Statistics

Were you
born in
Sweden?
Chi-Square a
3,489
df 1
Asymp.
Sig. ,062

67
a. 0 cells (0,0%) have
expected frequencies
less than 5. The
minimum expected cell
frequency is 37,1.

Figure 3.3 The variable age divided into classes with the expected age after using the Chi-square Goodness-of-fit test and
the target population statistics.

Age into classes

Observed Expected
N N Residual
18-21
years 18 19,9 -1,9

22-26
years 46 39,3 6,7
27-30
years 37 41,9 -4,9

Total 101

Figure 3.4 The variable ethnicity with the expected percentage after using the Chi-square Goodness-of-fit test and the
target population statistics.

Were you born in Sweden?

Observed Expected
N N Residual
Yes 75 65,9 9,1
No 28 37,1 -9,1
Total 103

68
8.4 In-depth tests (chapter 5)

Figure 4.1 Chi-square test in order to determine if there is a significant

difference between men and women in relation to feeling safe

Chi-Square Tests
Asymp.
Sig. (2- Exact Sig. Exact Sig.
Value df sided) (2-sided) (1-sided)
Pearson
a
Chi-Square ,833 1 ,361

Continuity
b
Correction ,512 1 ,474

Likelihood
,835 1 ,361
Ratio
Fisher's
,431 ,237
Exact Test
Linear-by-
Linear
,825 1 ,364
Association

N of Valid
103
Cases
a. 0 cells (0,0%) have expected count less than 5. The minimum
expected count is 23,31.
b. Computed only for a 2x2 table

Figure 4.2 Independent Samples Test in order to determine if there is a significant difference between men and
women in relation to safety behavior

Group Statistics
Std. Std. Error
What is your gender? N Mean Deviation Mean
Safety in Male 54 ,2917 ,47147 ,06416
relation
Female
to 49 ,1173 ,33697 ,04814
behavior

69
 Independent Samples Test
Levene's Test
for Equality of
Variances t-test for Equality of Means
95% Confidence
Interval of the
Difference
Sig. (2- Mean Std. Error
F Sig. t df tailed) Difference Difference Lower Upper
Safety in Equal
relation variances 4,854 ,030 2,139 101 ,035 ,17432 ,08149 ,01266 ,33598
to assumed
behavior Equal
variances
not 2,173 95,908 ,032 ,17432 ,08021 ,01510 ,33354
assumed

Figure 4.2.1 Reliability analysis to test the internal consistence of the different statements used to test against
gender

Reliability Statistics
Cronbach's N of
Alpha Items
,793 4

Figure 4.3 Chi-square test in order to determine if there is a significant difference between men and women in
relation to the reasons that contribute to feeling unsafe

Chi-Square Tests
Asymp.
Sig. (2-
Value df sided)
a
Pearson Chi-Square 10,779 5 ,056
Likelihood Ratio 12,527 5 ,028
Linear-by-Linear
4,150 1 ,042
Association
N of Valid Cases 102

70
Figure 4.4 Chi-square test in order to determine if there is a significant difference between men and women in
relation to avoiding narrow/small public spaces when it is dark outside

Chi-Square Tests
Asymp.
Sig. (2- Exact Sig. Exact Sig.
Value df sided) (2-sided) (1-sided)
Pearson
a
Chi-Square 4,928 1 ,026

Continuity
b
Correction 4,057 1 ,044

Likelihood
Ratio 4,961 1 ,026
Fisher's
Exact Test ,039 ,022
Linear-by-
Linear
Association 4,880 1 ,027

N of Valid
Cases 103

a. 0 cells (0,0%) have expected count less than 5. The minimum

expected count is 17,60.
b. Computed only for a 2x2 table

Figure 4.4.1 Cramér’s V test to test the strength of the relation determined with the chi-square test (figure 4.4)

Symmetric Measures
Approx.
Value Sig.
Nominal Phi -,219 ,026
by Cramer's V
Nominal ,219 ,026
N of Valid Cases 103

71
Figure 4.5 Spearman-Brown reliability test in order to determine if the two selected statements can be used
together

Reliability Statistics
Cronbach's Alpha Part 1 Value 1,000
N of a
1
Items
Part 2 Value 1,000
N of b
1
Items
Total N of Items 2
Correlation Between Forms ,329
Spearman-Brown Equal Length ,496
Coefficient Unequal Length
,496
Guttman Split-Half Coefficient ,474

Figure 4.6 Chi-square test in order to determine if there is a significant difference between men and women in
relation to use of public space & feeling safer during the summer than the winter

Chi-Square Tests
Asymp.
Sig. (2- Exact Sig. Exact Sig.
Value df sided) (2-sided) (1-sided)
Pearson Chi-Square a
3,948 1 ,047
b
Continuity Correction 3,152 1 ,076
Likelihood Ratio 4,010 1 ,045
Fisher's Exact Test ,058 ,037
Linear-by-Linear
3,909 1 ,048
Association
N of Valid Cases 103
a. 0 cells (0,0%) have expected count less than 5. The minimum expected count is
15,70.
b. Computed only for a 2x2 table

Figure 4.6.1 Cramér’s V test to test the strength of the relation determined with the chi-square test (figure 4.6)

Symmetric Measures
Approx.
Value Sig.
Nominal by Nominal Phi -,196 ,047
Cramer's
,196 ,047
V
N of Valid Cases 103

72
Figure 4.7 Independent Samples test in order to determine if there is a difference between the amount of time
lived in Stockholm and feeling safe in different seasons

Group Statistics
For how long have you lived in/near Std. Std. Error
Stockholm? N Mean Deviation Mean
I feel safer in public spaces during Between
29 ,66 1,203 ,223
the summer than during the 1-5 years
winter Longer
than 10 49 -,24 1,251 ,179
years

Independent Samples Test

Levene's Test
for Equality of
Variances t-test for Equality of Means
95% Confidence
Interval of the
Sig.
Difference
(2- Mean Std. Error
F Sig. t df tailed) Difference Difference Lower Upper
I feel safer in Equal
public spaces variances ,244 ,623 3,115 76 ,003 ,900 ,289 ,325 1,476
during the assumed
summer than Equal
during the variances
winter not 3,146 60,749 ,003 ,900 ,286 ,328 1,472
assumed

Figure 4.8 Chi-square test in order to determine if there is a difference between the visiting frequency of the
city in relation to avoiding public spaces when it is dark outside

Chi-Square Tests
Asymp.
Sig. (2-
Value df sided)
Pearson
a
Chi-Square 12,486 2 ,002

Likelihood
12,725 2 ,002
Ratio
Linear-by-
Linear
6,145 1 ,013
Association

N of Valid
102
Cases
a. 0 cells (0,0%) have expected count less than
5. The minimum expected count is 6,89.

73
Figure 4.8.1 Cramér’s V test to test the strength of the relation determined with the chi-square test (figure 4.8)
& Kendall’s tau-c test in order to determine the development of the variables

Symmetric Measures
Asymp.
Std. Approx. Approx.
a b
Value Error T Sig.
Nominal Phi ,350 ,002
by Cramer's V
Nominal ,350 ,002
Ordinal Kendall's
by tau-c ,282 ,096 2,921 ,003
Ordinal
N of Valid Cases 102
a. Not assuming the null hypothesis.
b. Using the asymptotic standard error assuming the null hypothesis.

74