Professional Documents
Culture Documents
Per-Olof H. Wikström Auth. Urban Crime, Criminals, and Victims The Swedish Experience in An Anglo-American Comparative Perspective
Per-Olof H. Wikström Auth. Urban Crime, Criminals, and Victims The Swedish Experience in An Anglo-American Comparative Perspective
Series Editors
Alfred Blumstein
David P. Farrington
Research in Criminology
With 49 Illustrations
Springer-Verlag
New York Berlin Heidelberg London
Paris Tokyo Hong Kong Barcelona
Per-Olof H. Wikstrom
Department of Criminology
University of Stockholm
S-106 91 Stockholm
Sweden
9 8 7 6 5 432 1
Acknowledgments
This book builds predominantly upon several different research projects on crime
in urban areas conducted during the 1980s and directed by the author. The vari-
ous projects have been carried out at the Department of Criminology, University
of Stockholm, and at the Research Unit of the National Crime Prevention Coun-
cil, Sweden. I wish to thank my colleagues at these two institutes for their en-
couragement, helpful comments, and assistance, and in particular my thanks
go to Professor Knut Sveri, Lars Dolmen, and Jan Ahlberg. Johan Gasste and
Kerstin Ohrnell have made a great contribution to some of the research presented
here in assisting with the coding and compilation of the data for the various parts
of what in this book is labeled the Stockholm 1982 Crime Survey (see Appendix
A).
Some of the findings presented in this book (parts of Chap. 3 and Chap. 7) are
a result of studies I have made working on Project Metropolitan, a longitudinal
study of a Stockholm cohort (see Wikstrom, 1985, pp. 117-133; 1987a, 1989a,
1989b, and Farrington & Wikstrom, 1990). Project Metropolitan is directed by
Professor Carl-Gunnar Janson, Department of Sociology, University of Stock-
holm, to whom I am grateful for valuable comments on my research. Some of
the findings, and some tables, in Chapter 3 have previously been published in
an article by the author in the Journal of Quantitative Criminology, Vol. 6. No
1, and are published in this book with the kind permission of Plenum Press.
The research reported in this book has been generously supported by grants
from many different foundations:
The Bank of Sweden Tercentenary Foundation.
The Swedish Research Council for the Humanities and the Social Sciences.
The Scandinavian Research Council for Criminology.
The National Council for Crime Prevention, Sweden.
The Swedish National Police Board.
and Coordination of Research, and was until 1985 supported by the Bank of
Sweden Tercentenary Foundation.
Dr. David P. Farrington has been my major source of inspiration in taking on
the task to complete this book. His encouragement has been invaluable to me.
This book was completed while I was a visiting fellow at the Institute of
Criminology, Cambridge University, England.
Contents
Acknowledgments............................................................... . .... v
vii
Vlll Contents
References. . . . . . . . . . . . . . . . . . . .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . .. . . . . . . . . . . . . . . . . . 246
Crime is largely an urban phenomenon, but the specifically urban and area dimen-
sions of the social processes that are connected with crime have been seriously
understated in much recent criminological work ... Such a claim could not have
been made forty years ago.
(Baldwin & Bottoms, 1976, p. 1).
The above statement by Baldwin and Bottoms about the neglect in crimi-
nology of the urban dimension of crime was made in the mid-1970s.
However, in the last decade there has been a significant upswing in theory
and research on crime in the urban environment. Also, new areas oftheory
and research into urban crime have come into focus. (For overviews see
Brantingham & Brantingham, 1984; Davidson, 1981.) One very good
example of the increasing interest in urban crime is the recent volume of
Crime and Justice entitled "Communities and Crime" (Reiss & Tonry,
1986), in which Reiss makes a strong argument for the importance of the
study of crime in urban communities and for the linking of the ecological
and individual traditions in theory and research on crime.
A review of the literature on crime in urban environments shows, not
unexpectedly, that Anglo-American research heavily dominates the scene
(Wikstrom, 1982; 1987b). Hence, much of the experience we have on
urban crime is based on North American and British research and theory.
In this book, I shall try to give the Swedish experience on urban crime in an
Anglo-American comparative perspective. The Anglo-American compara-
tive perspective means, first, that much of the analysis made in this book
takes as its point of departure the theories and findings of Anglo-American
research on urban crime and, second, that when possible, the Swedish
findings are compared with those of North American and British research.
Needless to say, this is not, nor is it intended to be, a comparative study.
But hopefully the results presented in this book might contribute interest-
ing questions for genuine cross-national research on urban crime. There is
a great lack of true cross-national comparative studies in criminology
(McClintock & Wikstrom, 1987), and it is my view that to advance our
1
2 1. Introduction
understanding of crime and its causes, there is a great need for the develop-
ment of such research. Cross-national studies are important for showing
us the common and unique features of crime in different countries, for
evaluating the generalizability of existing theories or for generating new
theories (Bennet, 1980), and for giving us perspectives on criminality in
our own countries and the policies and measures we use to handle it.
describe and analyze the offender characteristics and their patterns of crim-
inality. In Chapter 4 the focus is on the circumstances of crime, involving
victim and offender characteristics as an important part of these circum-
stances. Chapter 5 includes a special discussion of the urban victim and
some more analyses of the victim-offender relationship. Because this is a
book devoted to urban crime, a large section of it (Chapters 6, 7, and 8) will
deal with city crime patterns; among the issues taken up are the relation-
ships among housing, population composition, and offending, and the rela-
tionships among routine activities, area crime, and victimization patterns.
In Chapter 9, the conclusion, I shall summarize and try to link together the
findings from the previous chapters to give a tentative integrated picture of
urban crime, criminals, and victims.
1950 1.3
1955 1.5
1960 1.4
1965 l.2
1970 1.5
1975 1.6
1980 2.0
1981 2.5
1982 3.1
1983 3.5
1984 3.1
Source: Yearbook of Labour Statistics, 1976 (SCB)
and Statistical Abstract of Sweden, 1985.
Note. There have been some changes in the defini-
tion of unemployment over the period shown.
19 and under 26
20-29 14
30-39 16
40-49 11
50-59 11
60 and over 22
Source: Statistical abstract of Sweden, 1985.
Total 80 1128
TABLE 1.5. Traditional crimes recorded by the police in Sweden in 1950, 1960,
1970, 1980, and 1988 (per 10,000 inhabitants).
% increase
Crime 1950 1960 1970 1980 1988 1950 to 1988
aFor reason of comparability between years, violence toward a public official are not included
in the figures.
bCompleted homicides and assaults causing death; figures given with two decimal points.
c Attempted homicides, serious and petty assault.
dViolence toward a public official appeared first in 1965 as a separate category in the crime
statistics. The only important difference between the legal definitions of assaults and violence
to a public official is that the latter is restricted to assaults on public officials, which, according
to the law, should be regarded as a more serious crime than assaults on civilians.
eFigures given with two decimal points.
N.A. = Not available as separate category for this year.
6 1. Introduction
If one wants to reduce crime, one should, all other things equal, avoid to concen-
trate people in highly urbanized areas.
(Christie, 1975, p. 91, my translation)
The purpose of this chapter is to introduce the study of urban crime in
Sweden by considering its rate, structure, and trends.
The aggregate crime rate in any area or population may be viewed as a
result of the participation and frequency in crime, where participation is
defined as the percentage of individuals in the area or population ever
having committed a crime during the studied period, and frequency is the
average number of crimes per offender during that same period (see Blum-
stein, Cohen, Roth, & Visher, 1986, p. 20). The question about the parti-
cipation and frequency of crime in the urban population will be taken up in
Chapter 3.
When considering crime in an area, a further important divison may also
be made between crimes committed by residents of the area and crimes
committed by visitors to the area (see Table 2.1). In Chapter 7, it will be
shown that quite a big proportion of the crimes in the city of Stockholm-
more so for some than for other types of crime-were committed by
nonresidents. 1 Unfortunately, very little is known about how much of the
variation among different types of areas by degree of urbanization is due to
differences in crimes committed by residents and those committed by visi-
tors. A reasonable hypothesis, however, seems to be that the larger the
urban area, the more of its crime is committed by visitors. All in all, differ-
ences in crime rates between urban and other areas, between larger and
smaller urban areas, and so on, can be viewed as a result of differences in
participation in and/or frequency of crime by residents and visitors, and
changes in urban crime rates can be viewed as resulting from changes in the
participation in and/or frequency of crime by residents and visitors.
1. However, most of the nonresident offenders (71 %) were residents in the more
distant suburbs of the Greater Stockholm.
7
8 2. The Rate, Structure, and Trends in Urban Crime
TABLE 2.1. The area aggregate crime rate decomposed into its main components.
Residents Visitors
committing crimes committing crimes Total
Participation Pr P, PlOt
(% committing crime)
Frequency Fr F, F tot
(crimes per offender)
Crime rate CR r CRy CRtot
Crimes per
100.000
residents.
16000
14000
12000
..............
.,. ..
. ...... . .
10000
8000
6000
4000~~~~'-~~~~~-r-r-r-r~~~--~~~
50 75 100
Per cent living in urban area (locality)
TABLE 2.2. The rates per 10,000 inhabitants of police-reported traditional crimes in
the three big cities and in the rest of Sweden in 1988.
Greater Greater Greater Other
Crime Stockholm Gottborg Malmo Sweden
Assault" 67 53 61 41
Assault against female
by acquaintance 18 13 16 11
Assault against male
by a stranger 29 22 24 14
Completed homicides and
assaults causing deathb 0.24 0.22 0.26 0.14
Robbery 12 9 8 2
TABLE 2.3. The percentage of selected traditional crimes occurring in big-city areas
and in the rest of Sweden, and the percentage of population in the big cities and the
rest of the country in 1988.
TABLE 2.4. The rate per 10,000 inhabitants of police-reported assaultive crimes
(serious and petty assault, including homicides) by social context of crime in locali-
ties of different sizes.
Other circumstances 8 6 5 2
Total 63 39 33 10
two major victim surveys carried out in Sweden showed the same pattern
of difference between big-city areas and the rest of the country as that
shown by the police-reported crimes puts further confidence in the assump-
tion that there really are significant differences in the rate of crime related
to degree of urbanization (see Table 2.5). Each of the two victim surveys
included about 10,000 respondents. It should be noted that the victim sur-
veys only cover victimization of private persons victimizations and do not
include victimizations of public entities (e.g., companies). Moreover, it is
likely, as regards the violent crimes. that victim surveys and police re-
corded violence may capture different aspects of violence differently well;
it is likely that the police-recorded violence is a better representation of the
more serious violence that to a rather high degree involves abusers of alco-
hol and drugs and persistent criminals (see Chap. 3), whereas the victim
surveys are likely to best describe the violence the "ordinary" cittzen is
subjected to (for a discussion of this, see Wikstrom, 1985 pp. 36-55).
Trends in Crime
Comparing trends in crime by urbanization provides some evidence that
the general increase that seems to have taken place in most Western coun-
tries in the post war period has generally been much faster in urban as
compared to rural areas and biggest in very large conurbations (McClin-
12 2. The Rate, Structure, and Trends in Urban Crime
TABLE 2.5. Percentage of victims (aged 16-74) of traditional crimes in the big cities
and in other Sweden by size of muncipality.a
Violence Vandalism Theft
Source: 1978 figures: Levnadsforhiillanden 1978. Rapport 24, SCB. 1984185 figures: personal
communicatton from Lars Hall, Statistics Sweden.
'Violence resulting in visible marks or more serious injuries.
tock, 1981, p. 11). It was concluded in Chapter 1 that there has been a
significant upward trend for traditional crimes (violence, vandalism, and
theft) on the national level in Sweden in the postwar period. In this section,
the trends in police-reported traditional crimes for the period 1968 to 1988
will be compared for the three big cities and the rest of Sweden to see
whether their crime trends differ. There have been two major changes in
recording and statistics-gathering in Sweden-one in 1965 and one in
1968-both believed to have had an increasing effect on the crime rate; so
therefore, 1968 was chosen as the starting point for the trend comparisons.
The basic measure chosen was the average yearly linear change per
10,000 inhabitants (i.e., the regression coefficient). To be meaningful, such
an approach requires that the trends are linear in fashion, and in all cases
the visual inspection of the scattergrams and the calculated values of the
R2, justified a treatment of the trends as basically linear.
The use of the common measure of the percentage of change to compare
changes over time in the rate of crime between areas, was considered less
satisfactory, because it would mean measuring change relative to the rate
at the starting point of the comparison (in our case, 1968). Areas with high
rates at the starting point had to have a greater actual increase/decrease
than areas with a lower starting rate to end up with the same percentage of
change. The picture of the development of crime in big cities and other
parts of Sweden was also, in many cases, very different when the measure
of change was the percentage of increase/decrease.
The findings of the trend studies (see Table 2.6) show that (a) assaults
Explaining the Difference 13
TABLE 2.6. Crime rates per 10,000 inhabitants in 1968 and 1988, and regression
coefficient (and R2) for the period 1968 to 1988, for crimes of assault, vandalism
and theft, separately for the big cities and the rest of Sweden.
Crimes of theft
1968 727 592 518 325
1988 1217 967 1125 598
Vandalism
1968 43 28 36 36
1988 130 108 123 92
Assault
1968 34 27 34 17
1988 66 51 59 38
(serious and petty, including homicides), vandalism, and theft have in-
creased significantly over the studied period both in the big cities as well as
in the rest of Sweden, and (b) the level of increase for all three groups of
crime has been much faster in the big city areas than in the rest of Sweden.
This means that over the studied period (1968 to 1988), the difference be-
tween the big cities and the rest of Sweden in the rate of violence, vandal-
ism, and theft has increased significantly. For instance, in 1968, Greater
Stockholm had 17 assaultive crimes more per 10,000 inhabitants than the
non-big-city areas of Sweden, whereas in 1988, Greater Stockholm had 28
more assaultive crimes per 10,000 inhabitants than the non-big-city areas of
Sweden.
3. Routine activities are defined as "any recurrent and prevalent activities which
provide for basic population and individual needs" (Cohen & Felson, 1979, p. 593).
4. Direct-contact predatory crimes are defined as crimes in which someone takes or
damages the person or the property of another (Cohen & Felson, 1979, p. 589).
Explaining the Difference 15
The evidence from criminological research suggests that, at any given point
in time, the propensity to offend (i.e., break the law) varies significantly in
the urban population, ranging from those who occasionally may fall for an
extreme temptation (the pure opportunistic offender) to those who more
regularly search for opportunities to commit crimes (the chronic offender).
(See, e.g., Leblanc & Frechette, 1989, pp. 181-182). As to violent be-
havior, the same appears largely to hold true, although it is perhaps better
to speak about a differential readiness to react with violent actions in differ-
ent kinds of situations. The propensity to offend shows a general variation
with demographic and social characteristics such as sex, age, social class,
and so on, as most textbooks in criminology will report (e.g., Nettler, 1974
pp.98-135).
Recently, there has been a strong emphasis on the value of a develop-
mental perspective on offending, acknowledging that an individual's cri-
minal propensity is something that may grow and change in strength over
time but that may also diminish, and that the factors that influence offend-
ing at different stages in a process of drifting in or drifting out of a "crimin-
al life-style" may vary (Farrington, 1988a; LeBlanc & Frechette, 1989;
Loeber & LeBlanc, 1990). Farrington has summarized this perspective by
saying that:
It is plausible to propose that behavior is a function of the interaction between the
person and the situation. Immediate situational causes of offending can be distin-
guished from long-term causes. Individual differences in criminal propensity, built
up over time in a learning process, may interact with immediate situational factors
such as opportunities, costs, and benefits, so that the potential becomes the actual-
ity (the criminal event). (pp. 319-320)
16
Offenders in Urban Crime 17
Sex
The commission of traditional crimes is predominantly a male business. It
is well established that females have a very low involvement in traditional
crimes. The female involvement tends to be higher in less serious crimes.
The highest female involvement of those crimes included in the Stockholm
1982 Crime Survey is in molestation, followed by violence toward a public
official, serious assault, and vandalism, (see Table 3.1). In this context, it
should be pointed out that the cases of violence toward a public official
generally were of a less serious nature when involving a female than a male
offender. Moreover, the fact that the female involvement in police-
recorded serious violence is higher than for the female involvement in
petty assaults is likely to be related to two circumstances: First, many of
the cases of serious violence with female offenders can be described as
"defense situations" where the female was under, or previously had been
under, attack from the (almost always) male victim, and was defending
herself against a physically superior person, often using a weapon, which
generally means more severe injuries. Second, one may speculate whether
the reportability of male victims in less serious cases of assault by females
may be lesser than the corresponding reportability for female victims.
That there is an overall great difference in the rate of offending between
males and females is well-known. However, there are also significant dif-
ferences in the patterns of offending between males and females, an issue
that has not received the same attention and is discussed in the section
below on offending and its development.
18 3. The Urban Offender
TABLE 3.1. The sex distribution of offenders by different legal categories of crime
(the city of Stockholm, 1982).
Males Females
Crime (%) (%)
Serious assault a 90 10
Petty assault 94 6
Violence towards a public official 83 17
Robberies against the person 98 2
Rape 100 0
Molestation 81 19
Vandalism 91 9
Residential burglary 93 7
Nonresidential burglary 95 5
Thefts of cars 95 5
Thefts from cars 95 5
Note. The legal definitions of the crime categories are given in Appendix B.
a Including homicide.
Age
Another well-established difference in offending is its variation by age. The
age distributions for the different crimes included in the Stockholm 1982
Crime Survey is shown in Table 3.2. When comparing the age distributions
of different categories of crime, it should be noted that it is valid only to
compare the pattern of age variation; comparison among crime categories
of the rate at given ages is problematic due to the great differences in dark
figures and clearance rates. In general, crimes against the person have a
higher dark figure and higher clearance rate compared to crimes of theft
and vandalism (see Persson, 1980; Wikstrom, 1987b).
The results presented in Table 3.2 show significant variations in the age
distribution of offenders among the different categories of crime. Robbery,
vandalism, nonresidential burglary, and thefts of cars have a peak for
offenders aged 15 to 19. Violence toward a public official and residential
burglary have a peak in the next age group, 20 to 24. Finally, serious
assault, rape, molestation, and thefts from cars peak for offenders at higher
ages, 30 to 34. Petty assault does not have any significant peak ages, but
rather a peak level for offenders from the ages of 15 to 44.
It is important to point out in this context that the social contexts of the
crimes (to be further discussed in Chapter 4) vary significantly with age for
crimes of assault and vandalism. Although the great majority (around two
thirds) of the assaults (serious and petty) committed by offenders below the
age of 20 are committed in public places (streets, squares, public transport,
and places of public entertainment), this proportion continuously de-
TABLE 3.2. The age distribution of offenders in different legal categories of crime per 10,000 in respective age group; peak ages in bold type
(the city of Stockholm, 1982).a
Age Serious Petty Violence toward Robbery Rape Molestation Vandalism Residential Nonresidential Thefts of Thefts from
assaultb assault public official against person burglary burglary cars cars
10-14 0.0 11.7 0.0 0.0 0.4 4.7 23.7 0.0 23.7 0.0 0.0
15-19 2.2 86.2 25.4 32.9 2.2 16.0 119.2 13.2 248.4 53.3 51.7
20-24 3.9 89.4 42.5 15.8 2.7 20.3 69.8 132.7 171.3 45.2 47.2
25-29 6.7 91.0 30.9 13.2 2.1 26.3 60.1 59.9 157.7 26.6 45.3
30-34 7.6 85.3 27.3 10.6 3.8 31.8 49.3 32.2 140.0 23.7 75.1
35-39 7.5 90.3 21.9 13.6 3.2 18.3 50.8 58.7 68.2 21.3 20.1
40-44 6.4 91.8 14.0 4.1 2.6 27.1 26.2 27.1 57.7 14.6 19.2
45-49 4.1 49.6 12.3 2.4 1.0 17.4 41.0 20.1 24.6 6.8 0.0 0
50-54 4.1 19.1 4.4 0.9 0.6 16.8 2.9 12.3 17.6 0.0 9.7 l
::l
55-59 0.0 22.6 2.8 0.0 0.7 7.1 4.7 1.9 1.4 1.2 0.0 0..
(1)
60 and above 0.0 6.2 0.8 0.0 0.1 4.8 1.1 0.0 0.3 0.0 0.0 ....
'"5'
a Including
homicides. C
....
bOnly offenders residing in Stockholm included. cr"
I'l
::l
n
::to
S
(1)
......
'-D
20 3. The Urban Offender
creases with age. At ages 20 to 29 just about one half of the offenders
committed assaults in public places. At ages 30 to 39 and above, a clear
majority of the offenders committed assaults in nonpublic space. For those
aged 40 and older, the majority of assaults were committed within the
family or among acquaintances in residences. Similar age-related differ-
ences can be observed for vandalism, where just above two thirds of the
cases occur in public for offenders below the age of 20, just below one half
for offenders aged 20 to 29, and clearly less than one half for offenders aged
30 and older. The relevance of pointing this out is to say that assault and
vandalism in public show a different age pattern than that for assaults and
vandalism in private and semiprivate spaces.
Ethnicity
Of the population of the city of Stockholm during 1982, 8% were foreign
citizens. In the age group 16 to 44, this proportion was slightly higher (i.e.,
13%). However, citizenship is a less-than-perfect measure of ethnicity be-
cause many first-generation immigrants have acquired Swedish citizenship,
and some people with foreign parents (second-generation immigrants) may
have kept their foreign citizenship. Therefore, a comparison of offenders
by citizenship, such as the one that will be conducted in this section, has its
drawbacks. It would have been best to have the possibility to compare the
patterns of offending for the groups of first-generation immigrants, second-
generation immigrants, and the rest ofthe population. This was, however,
unfortunately not possible within the scope of this study.
The greatest number of residents with foreign citizenship in Stockholm
are of Finnish origin, whereas the countries from which most of the Stock-
holm residents with a non-Nordic foreign citizenship come are Greece,
Turkey, and Yugoslavia. In the comparison of offenders by ethnicity, I
shall show this for three categories: Swedish, other Nordic, and non-
Nordic. It was not judged possible to have a finer divison because of the
generally small number of offenders for many nationalities.
The findings (see Table 3.3) show that those with a foreign citizenship,
given that they are 3% (Nordic) and 5% (non-Nordic) of the Stockholm
residents, are clearly overepresented among the offenders, with one signif-
icant exception, and that is for non-Nordic citizens' involvement in serious
property crimes where they, by and large, are an expected proportion of
the offenders, perhaps even a somewhat lesser proportion than expected,
because there is a higher percentage of non-Nordic foreigners among the
population in the most crime-active ages. The highest proportion of foreign
offenders is found for crimes of assault (serious and petty) and especially
rape. An alternative way to say this is that these are the crimes with the
lowest percentage of Swedish offenders because, without having any point
of reference, it is really difficult in comparisons of this type to say whether
it is the one category that has a high level or the other that has a low level.
Offenders in Urban Crime 21
Serious assault b 75 14 11
Petty assault 73 9 18
Violence toward a public official 77 13 10
Robberies against the person 84 6 10
Rape 50 17 33
Molestation 80 6 14
Vandalism 80 8 12
Residential burglary 91 7 2
Nonresidential burglary 84 11 5
Thefts of cars 86 11 3
Thefts from cars 87 5 8
Note. The legal definitions of the crime categories are given in Appendix B .
Only offenders residing in Stockholm included.
blncluding homicide.
TABLE 3.4. The previously recorded criminality (all kinds) by extent for offenders
by different legal categories of crime (the city of Stockholm, 1982).
Extent of previous crime
Serious assault" 31 27 42
Petty assault 44 29 27
Violence toward a public official 32 24 44
Robbery against the person 29 24 47
Rape 39 34 27
Molestation 69 14 17
Vandalism 39 31 30
Residential burglary 7 13 80
Nonresidential burglary 27 25 49
Thefts of cars 18 19 63
Thefts from cars 33 24 43
a Induding homicides.
son, 1969 pp. 218-222; Olofsson, 1971 pp. 146-151; Werner, 1971; SOU,
1976; Tham, 1978 pp. 130-136; Janson, 1982).
Perhaps more striking than the dominance of offenders belonging to low-
er social class backgrounds, when looking at an urban offender population
cross-sectionally, is the very high degree of previous records for crime and
the extent of such records among the offenders. In the Stockholm 1982
Crime Survey, offenders were classified according to their previous crimin-
al record (for all kinds of crimes) as follows:
1. None (i.e., no previous record); the crime included in the survey was
the offenders first recording.
2. Recidivists (i.e., those with a previous record but not qualifying for
being classified as persistent-(see below).
3. Persistent (i.e., those with at least one recording for crime in at least
four different years previous to 1982).
The majority of the offenders, with the exception those who committed
molestation, have a previous record for crime, ranging from 56% in petty
assaults to 93% in residential burglaries .
For many of the crime categories, in particular for crimes of serious theft,
robbery, violence toward a public official, and serious assault, the pro-
portion of offenders with an extensive previous record is very high,
ranging from 42% for serious assault to 80% for residential burglaries.
Offenders in Urban Crime 23
TABLE 3.5. Extent of previous recordings for crime (all kinds) in a 1953 Greater
Stockholm cohort (Project Metropolitan) at the age of 25/26 (June 1979).
Recidivisfs a
None (including one-timers) Persistent
(%) (%) (%)
a Note that all offenders (even one-timers) are included in this table in the recidivist category;
otherwise, the categones are defined the same way as in Table 3.4.
The picture is clearly one of the urban offender as a person with a heavy
criminal involvement, at least as regards the somewhat more serious tradi-
tional crimes. If one compares the proportion of offenders previously re-
corded with what could be expected from a normal population (i.e., the
amount of recordings in the 1953 Greater Stockholm cohort for offenders
up to the age of 25 1) the difference is dramatic (see Table 3.5).
For 5 out of the 11 categories of crime it was possible in the Stockholm
1982 Crime Survey to extract information about whether the offender was
known as an intravenous drug addict or not (see Appendix A for more
information on this variable). For all crime categories, information was
available about whether or not the offender was previously recorded for a
drug offense. Because there was a rather close correspondence between
the level of offenders being known as intravenous drug addicts and being
recorded for previous drug offenses, the latter may be taken as an indica-
tion of the proportion of drug abusers in the crime categories where data
about intravenous drug abuse is lacking. It should be mentioned in this
context that a state commission, using a so-called snowballing technique,
tried to estimate the number of intravenous drug abusers in the Greater
Stockholm area in 1979. The commission estimated that the intravenous
drug abusers consisted of less than 1% of the total Greater Stockholm
residential population (Ds S, 1980, p. 92).
Clearly adding to the picture of urban offenders in more serious tradi-
tional crimes as socially marginalized ("habitual" criminals and/or abusers
of drugs and alcohol) is the fact that a very high proportion of them, espe-
cially in crimes of gain (robbery and theft), are known drug addicts (Table
3.6). Although no measure of alcohol abuse was included in the Stockholm
1982 Crime Survey, it is well-known from previous Swedish studies, at least
with regard to the somewhat older violent offenders, that to a rather high
degree they are known for alcohol abuse (Wikstrom, 1980; Kuhlhorn,
1984, pp. 142-146).
1. Judging on the basis of the pattern of decay with age in first-timers (see further
below) in the cohort, the proportion that will debut in crime after the age of 25 can
be estimated to around 4 to 5%.
24 3. The Urban Offender
TABLE 3.6. Percentage of offenders known for intravenous drug abuse,a and per-
centage of offenders with a previous record of drug offenses (the city of Stockholm,
1982).
Known intravenous Previous record for
Crime drug abuse (%) drug offenses (%)
Serious assaultb 13 23
Petty assault 6 8
Violence toward a public official N.A. 17
Robbery against the person 26 27
Rape N.A. 10
Molestation N.A. 9
Vandalism 8 13
Residential burglary 49 68
Nonresidential burglary N.A. 35
Theft of cars N.A. 44
Thert from cars N.A. 49
"Data about the offenders' intravenous drug use was only collected for some of the included
crimes.
blncluding homicides.
N.A. = Not available in study.
2. The findings presented here are predominantly taken from several studies made
by the author while working on Project Metropolitan (see Wikstrom, 1985, pp.
117-133; Wikstrom, 1987a; Wikstrom, 1989a; Wikstrom, 1989b; Farrington &
Wikstrom, 1990; Wikstrom, 1990).
Crime and Its Development in the Urban Population 25
of 13 and 25. Some results, however, will cover the period up to the middle
of 1979, the year during which the cohort members reached the age of 26.
The number of crimes recorded for the cohort members is calculated in
total on an annual basis and divided into seven subclasses: theft, fraud,
vandalism, violent crimes, traffic crimes, narcotic crimes, and other crimes.
It should be pointed out that in this classification, robbery has been
counted as a violent crime; that receiving of stolen goods is included in
theft; that arson is included under vandalism; and, finally, that the category
of "other crimes" is very heterogeneous, including such crimes as defama-
tion and crimes against the Taxation Offense Act. (For more detailed in-
formation about the contents ofthe classes of crime, see Wikstrom, 1990.)
TABLE 3.7. Participation and frequency in crime for the Greater Stockholm
Cohort, ages 13 to 25/26 (from 1966 to June, 1979) by sex and social class of
offender.
Subgroup of offenders or type of crime Participation (%) Frequency
Total 19 8.1
Males 31 9.1
Females 6 4.5
Theft 11 7.0
Fraud 3 3.3
Vandalism 3 1.6
Violent crimes 4 2.2
Traffic crimes 8 3.5
Narcotic crimes 2 3.7
Other crimes 7 1.9
counterparts. The frequency is twice as high for males (9.1) as for females
(4.5).
The social class of the cohort members parents was measured at two
points in time; at the cohort members' birth (1953) and at their age of 10
(1963). In the present analysis, only data about the parents' social class at
the cohort members' age of 10 wlll be used. A fourfold classification will
be used, dividing the parents' social class into lower working class
(N = 2,416), working class (N = 3,306), lower middle class (N = 6,396) and
upper middle class (N = 2,587)-(see Wikstrom, 1987a, pp. 15-17). Since
the upper class was such a tiny part of the highest soclal class, it was in-
cluded in the upper middle class. The basis for the social class classification
is the type of occupation of the head of the cohort member's family, gener-
ally the father. For 412 of the 15,117 cohort members, or 3%, the head of
the family had no occupation, or information on occupation was lacking.
The criminality of those cohort members belonging to this group is higher
than for those whose parents social class was known.
The participation rate varies significantly among the social classes. The
participation rate is more than twice as high for the lower working class
than for the upper middle class. Not only is the participation in crime less
for higher social class members, but they are also less active in crime, as the
calculation of the frequency in crime for offenders from different social
classes shows-the frequency for the lower class offenders is twice as high
as that for the upper-middle-class offenders.
The Distribution of Crime in the Urban Population 27
TABLE 3.8. The skewness in the offender distribution as measured by alpha (Gini
coefficients ).
Alpha N
offenders) and 1 if the distribution of the crimes among the cohort mem-
bers (or offenders) is maximally unequal. Fox and Tracy claim this to be a
better measure of the crime distribution than the commonly used measure
of X% of the offenders as being responsible for X% of the crimes. A major
advantage with alpha, according to Fox & Tracy, is that "regardless of the
size of cohorts being studied or differences in the concentration of delin-
quents or the total number of offenses that they have committed, the index
can be used to compare the offense distributions observed" (pp. 272-273).
They get alphas for the full male cohorts of .82 (1945) and .84 (1958) and,
for offenders only, of .47 (1945) and .51 (1958) (pp. 266-267).
The same exercise was carried out for the Greater Stockholm cohort.
The alpha (Gini coefficient) for the Greater Stockholm cohort was .94 for
the total cohort and .71 for offenders only. The alpha was also calculated
separately for the two sexes, for males, excluding the traffic crimes, and
the crimes belonging to the heterogeneous category of other crime. (see
Table 3.8). There were no dramatic differences in the value of the alphas
among the different categories for which it was calculated. The result indi-
cates a stronger concentration of crimes in the Stockholm cohort than in
the Philadelphia cohorts. However, one should be careful before jumping
to conclusions, because, as stated by Fox and Tracy (1988),
The obtained alpha will depend on whether the sample consists of only offenders or
of both offenders and nonoffenders, on whether all types of offenses are included or
just serious or violent crimes, and on whether or not the sample is restricted to a
particular age range. (p. 272)
It cannot be excluded that the Philadelphia cohorts and the Stockholm
cohort may show some differences as regards the type of factors mentioned
by Fox and Tracy (1988) that may disturb a direct comparison. The Lorenz
curves for the total male cohort and the male offenders only are shown in
The Distribution of Crime in the Urban Population 29
100
90
80
.,"'
E
.;: 70
<J
., 60
C
.,...
<J
50
.,
Cl.
...
.~
..!!!
40
::::J
E 30
::::J
U
20
10
10 20 30 40 50 60 70 80 90 100
Cumulative per cent male cohort members
FIGURE 3.1. The Lorenz curve for the distribution of crime for the full Greater
Stockholm male cohort (N = 7,719).
100
90
80
.,"'
. 70
....,c:
<J
60
.,...
<J
50
.,
Cl.
>
.;::; 40
..!!!
::::J
E 30
::::J
U
20
10
10 20 30 40 50 60 70 80 90 100
Cumulative per cent male offenders
FIGURE 3.2. The Lorenz curve for the distribution of crime for male offenders only
in the Greater Stockholm cohort (N = 2,389).
30 3. The Urban Offender
Figures 3.1 and 3.2, respectively. The Lorenz curves gives more detailed
information on what kind of distribution lay behind the Gini coefficient (or
alpha). Following the praxis of Fox and Tracy, the normal way of present-
ing the Lorenz curve has been mirrored.
Impressive as the figure of the distribution of crimes by cohort members
is, it is not, however, without fault. It is likely that low-rate offenders are
less represented among the recorded offenders than frequent offenders,
which contributes to an overestimation of the skewness of the distribution
of crimes by cohort members (Wikstrom, 1987a, pp. 4-8). Further, but of
less significance, the figures are calculated on the number of crimes each
offender is known for; obviously, some of these crimes might have involved
several cohort members, meaning that some crimes might have been
counted more than once. All in all, it seems unlikely that the bias pointed
to would alter the main conclusion that a minority of the offenders account
for a great part of the cohort members' criminality.
'the discovery of the skewed distribution of crimes by offenders has led
to an interest in the so-called chronic (or persistent) offender, that is, the
offender belonging to the small group of offenders who commit a large part
of the crimes, although it has not been easy to establish any criteria for
which offenders should be counted as chronic and which should be not
(see, e.g., Wolfgang et aI., 1972, Chap. 6; Blumstein et aI., 1985). It has
been argued that the early prevention of the offenders from developing
into "chronics" could substantially reduce crime. As was shown in the pre-
vious section, the cross-sectional picture of the majority of urban offenders
in more serious, traditional crimes was one of a chronic offender.
These studies strongly suggest, then that varieties of delinquency are not indepen-
dent, but more or less consistent with each other. Thus, there is nearly universial
support among the studies examined for the conclusion that at least one dimension,
perhaps loosely configured, of delinquency exists. (pp. 67-70)
Specialization and Versatility in Crime 31
The choice of the type of the next offense is only very slightly related to the type of
prior offense or offenses. This finding leads us to the conclusion that the next type
of offense-be it injury, theft, damage, combination, or nonindex-cannot very
32 3. The Urban Offender
well be predicted by examination of the prior offense history, at least when the
history is represented by our typology. There is practically no evidence to support a
hypothesis of the existence of offense specialization among juvenile delinquents.
(p.254)
A U.S. study of 1,619 juvenile offenders aged 8 to 17 in Pima County,
(Arizona), which was carried out by the same method (transition probabil-
ity matrix) as that of Wolfgang et aI., reaches the same conclusion about
specialization: "There was no evidence that male or female juveniles
specialize in a particular type of offense as their official career develops,
although the nature of their involvement differs" (Rojek & Erickson, 1982,
p.26).
In studying specialization and versatility of offenders in the 1953 Greater
Stockholm cohort, I shall rely upon an analysis of patterns of correlations
and, as I have termed it, the specificity ratio. One might say that the corre-
lational analysis attempts to study the existence of patterns of versatility,
whereas the calculation of the specificity ratio focuses upon specialization
in crime.
TABLE 3.9. The specificity ratio by type of crime (only offenders known for three or
more crimes included).
Offenders known for:
Percent
crimes Theft Fraud Traffic Narcotic Violent Vandalism
0-25% 24 63 48 70 73 84
26-50% 28 19 34 21 20 13
51-75% 25 7 13 5 5 3
76-100% 23 11 5 3 2
Note. The same offender may appear in several columns, although, for obvious reasons, an
offender can only appear with 51 % or more of his crimes in one specific crime category.
TABLE 3.10. The specificity ratio by type of crime and sex of offender; only offen-
ders known for three or more crimes (figures in percentages).
0-25% 24 21 72 23 38 32 73 55 74 54 83 81
26-50% 29 25 16 36 42 50 19 35 20 26 13 14
51-75% 25 24 6 10 13 14 6 0 4 18 3 0
76-100% 22 30 6 30 7 4 2 9 2 3 5
Note. For the female offenders in the cases of traffic crimes, vandalism, and violent crimes the
percentage is calculated on 20-40 offenders.
Theft 39 44 14 40 18 18
Violent 39 13 29 16 26
Vandalism 8 17 13 16
Fraud 16 9 13
Traffic 16 53
Narcotic 33
Note. Coefficients multiplied by 100 (N = 2,837); coefficients 30 and higher in bold type.
Theft 79 33 21 63
Violent crimes 73 34 17 54
Vandalism 81 14 4 67
Fraud 15 16 98 95
Traffic crimes 41 74 31 59
Narcotic crimes 16 64 -2 43
Other crimes 24 86 19 74
Contribution 36 16 13
criminality (.33), followed by modern criminality and fraud (.22) and tradi-
tional criminality and fraud (.18).
Theft 46 44 64 59
Violent crimes 7 81 17 67
Vandalism 15 67 1 46
Fraud 9 0 85 76
Traffic crimes 89 12 22 79
Narcotic crimes 64 20 46 51
Other crimes 61 12 7 85
Contribution 35 17 14
Age Patterns
So far, offending in the 1953 Greater Stockholm cohort has only been de-
scribed by its members' participation in and frequency of crime, by the
distribution of crime in the cohort, and by the degree of specialization and
versatility of crime in the cohort. Now it is time to consider the develop-
ment of offending in the cohort against this background.
It is well-known that criminality is not evenly distributed for different
ages (Sveri, 1961, Farrington, 1986b). Several studies show that the peak
age for criminality tends to be somewhere in the age range of 15 to 17.
Farrington (1984) has shown that "self-reported and official delinquency
tend to peak at about the same age" (p. 253).
There are, however, differences among types of crime with respect to the
age distribution of offenders. Violent crimes peak later than property
crimes. Hirschi and Gottfredson (1983) point out that this pattern is well-
established for official data but not for self-report data. They interpret this
as a result of the fact that seriousness of violence is age-related, unlike the
seriousness of property offenses and state that" 'injury' offenses by the
very young are unlikely to be sufficiently serious to attract the attention of
officials" (p. 558).
The peak age range for total crimes in the Stockholm cohort is from 15 to
17, with the highest incidence at the age of 17. However, the peak ages for
different types of crime vary significantly (see table 3.14); for example, the
peak age for crimes of theft and vandalism is 15, the peak age for fraud is
from 21 to 24 (with the highest incidence at age 23), the peak age for
violence is 19, and the peak age for other crimes is 20.
Traffic crimes and narcotic crimes cannot be said to have any distinct
peak age; the highest levels of narcotic crimes are found between age 20
and 23, and traffic crimes have a slightly fiuctating level of around 360-400
crimes per year between the ages of 16 and 25, with peaks at ages 16 and
25, and a dip at age 21.
Age Patterns 37
TABLE 3.14. Number of recorded crimes for the 1953 Greater Stockholm cohort.
Age Theft Fraud Vandalism Violence Narcotic Traffic Other Total
13 862 15 73 10 2 0 10 972
14 1,414 24 45 13 13 1 18 1,528
15 1,899 49 83 43 119 136 163 2,492
16 1,587 67 67 88 49 405 177 2,440
17 1,624 80 71 151 82 361 297 2,666
18 1,039 88 49 105 117 377 285 2,060
19 727 180 39 178 89 387 245 1,845
20 610 208 49 134 124 370 364 1,859
21 443 242 47 92 122 338 314 1,598
22 419 233 38 105 118 366 318 1,597
23 373 256 30 130 129 364 207 1,489
24 434 222 52 93 93 394 199 1,487
25 390 142 55 93 81 408 76 1,245
300
250
200
Crimes
150
Offenders
50
0
15 20 25
Age
FIGURE 3.15. Number of total offenders and total crimes of the 1953 Greater Stock-
holm cohort as an index (index age 13 is 100).
Sex Differences
Looking at the age-specific crime rate and crime per offender rate for males
and females reveals significant differences. The male crime rate is highest
in the ages 15 to 17, which is also the period with the highest crime-per-
offender rate for males. The female crime rate is highest in the ages of 22
to 24, which is also the period with highest crime-per-offender rate for
females (Wikstrom, 1990).
As regards differences in the male and female crime rate, the difference
is largest at the ages of 16 to 17, when the male rate is 23 to 25 times higher
than the female rate, as contrasted to the oldest of the compared ages,
when the male rate is around 5 times higher than the female rate. Also, the
difference in crime-per-offender rate is highest at the youngest ages,
whereas there is no, or practically no, difference in the crime-per-offender
rate at the oldest ages (Wikstrom, 1990).
rate among the social classes tend to decrease with age. As regards the
age-specific crime-per-offender rate, all social classes have a peak at age
15, except the working class, which has a peak at age 16 (Wikstrom, 1990).
Looking at the offender rates, the middle-class offender rates tend to
peak later than the working-class offender rates. The two working classes
have a peak in offender rates at the age of 17, whereas the middle-class
offender rates tend to peak around the age of 20 (Wikstrom, 1990).
tribution of career and noncareer criminals. Of all offenders, 37% are one-
timers, and as much as 63% have committed no more than three crimes.
It has already been shown that with increasing age there is an increasing
proportion of recidivist offenders. Looking at the crimes committed at each
age, the increasing importance with age of the recidivist offenders for the
criminality of the cohort is even more striking. At age 15, as many as 61 %
of crimes are committed by recidivist offenders, and at age 25 the recidivist
offenders account for 92 % of the crimes (see Table 3.17.) The fact that the
recidivists account for a higher proportion of the crimes than their pro-
portion of the offenders indicates that recidivist offenders have a higher
lambda than first-timers.
Sex Differences
For both males and females, the rate of first-time offenders shows a con-
tinuous decrease, and the rate of recidivist offenders shows a continuous
increase over the ages studied. However, the decrease is much faster for
the males than for the females. There are nine times more male beginners
at age 13 than at age 25. The corresponding figure for females is three times
more. In contrast, the increase in recidivist offenders is faster for females
than for males. There are five times more female recidvist offenders at age
25 than at age 14. The corresponding male figure is two and one half times
more (see Table 3.18).
42 3. The Urban Offender
Source: "Age and Crime in a Stockholm Cohort" by P.-O. Wikstrom, 1990, Journal of Quan-
titative Criminology, 6.
TABLE 3.18. Offenders, first-time offenders, and recidivist offenders by sex per
1,000 males respective to females in the 1953 Stockholm cohort.
Males: Females:
first-time Recidivist first-time Recidivist
Age Offenders offenders offenders Offenders offenders offenders
13 51 51 0 8 8 0
14 48 35 13 6 6 0
15 52 29 23 6 5 1
16 65 34 31 6 4 4
17 67 30 37 8 5 3
18 65 24 41 7 5 2
19 66 23 43 9 6 3
20 64 20 44 8 4 4
21 63 20 43 8 4 4
22 55 16 39 9 4 5
23 46 9 37 8 3 5
24 42 8 34 8 3 5
25 39 6 33 8 3 5
Source: "Age and Crime in a Stockholm Cohort" by P.-O. Wikstrom, 1990, Journal of Quan-
titative Criminology, 6.
Age Patterns 43
At every age, there are more male than female offenders, both first-time
offenders and recidivist offenders. However, the gap between males and
females is less at the oldest of the compared ages. There seems to be a peak
in the male/female rate differences around the ages 15 to 17, thereafter
there is more or less a continuously decreasing sex difference.
TABLE 3.19. Crimes per offender in the 1953 Greater Stockholm cohort.
Age Theft Fraud Vandalism Violence Narcotic Traffic Other Total
The results of the present study show that frequency (lambda) in the
Stockholm cohort does vary with age and, further, that the variations are
different for different types of crime.
The calculation of frequency (i.e., the age-specific crime-per-offender
rates) reveals higher rates for total crimes for offenders aged 15 to 17, with a
peak at age 15 (see Table 3.19). As regards specific types of crimes, there is
an increase in the crime-per-offender rate for theft, with a peak at age 15
and thereafter a decrease in rate to the age of 21, where the rate seems to
stabilize at around 2.5 crimes per offender. Fraud and traffic crimes have a
continuously increasing crime-per-offender rate over the ages with a signif-
icant jump upward for fraud at age 19 and for traffic crimes at age 15 and
at age 24. Also, the crimes-per-offender rate for violent crimes might be
described as continuously increasing with age, but the increase is rather
small, from around 1.2 crimes to around 1.5 crimes per offender. Vandal-
ism has generally the highest crimes-per-offender rate for the youngest of
the studied ages. Narcotic crimes have, after an increase, a peak at age 15
. and thereafter a decrease followed by a jump upward, stabilizing at a rate
around two crimes per offender at age 20. The heterogeneous category of
other crimes shows more irregular rates than any of the specific crime
categories.
In Blumstein et al. (1986), it is argued that "if active in a crime type,
females commit that crime at rates similar to those of active males" (p. 67).
The results of the present study contradict to that statement, because they
show that lambda in the Stockholm cohort is not invariant by sex. Further,
in the present study the lambda also shows significant differences among
offenders from different social classes.
Discussion 45
Discussion
According to Blumstein et al. (1986), "A criminal career is the character-
ization of the longitudinal sequence of crimes committed by an individual
offender" (p. 12). They go on to say that the criminal career approach
"partions the aggregate rate into two primary components: participation,
the distinction between those who commit crime and those who do not,
and frequency, the rate of activity of active offenders." Changes in aggre-
gate crime rates may reflect changes in (1) the level of participation, (2)
individual frequency, and (3) duration of activity. (p. 13). It is largely
beyond dispute that the overall total crime rate peaks in the youth period
and thereafter declines, although closer examination shows that there
might be significant variations among times, places, and types of crimes
and for demographic subgroups.
other types of crime show an increase over the ages. There are 18.5 times
more male crimes at age 13 to 16, compared with 6 times more male crimes
at age 21 to 25/26. Consequently, the proportion of all crimes committed
by females increases with age but never reaches as high as one fifth.
Age at onset tends to be higher for females than for males. Although
there is a great difference in the number of female and male first-time
offenders at age 13 (6 times more males), the difference at age 25 is signif-
icantly smaller (2 times more males). In contrast to the males, females
with a late start tend to be more persistent than those with an early age at
onset. The female and male crime structure tends to become more similar
with increasing age.
All in all, male and female criminality tend to narrow in many respects
with increasing age. Another way to put it is to say that it is in youth that
the differences between male and female criminality are most pronounced.
One hypothesis to account for this might be that females are more pro-
tected from criminal influences during the most criminogenic period, or
that youth is not criminogenic in the same way for females as it is for males.
That might depend on such factors as closer parental control and fewer
group activities among females in youth. Much crime in youth is connected
with group activities (Sveri, 1965), and poor parental control is one major
correlate with criminality found in other studies (e.g., Farrington, 1984).
In this context, it can be noted that the crime with the greatest female
involvement is fraud, mainly a nongroup kind of crime with a rather high
median age at onset. It might also be that the revealed pattern of narrow-
ing has something do to with changes in the female position in society
during the period studied.
The social class differences in crime are much greater for crime rates
than for offender rates, indicating not only a higher proportion of offenders
in lower classes but also a higher activity in crime, although the small pro-
portion of the offenders who account for one half of the crimes is about the
same in the different social classes. The social class differences in crime
rates are greater for traditional crimes, in particular for violent crimes,
than for modern crimes. There are no marked differences in the degree of
specialization in crime among offenders from different social classes. It is
only the upper-middle class offenders who have a deviating crime struc-
ture.
Working-class crime develops differently from middle-class crime. The
latter increases over all age periods, whereas the former drops off in the
last age group. The social class differences in incidence of crime narrow
with age.
Age at onset tends to increase with the social status of the offender. The
difference among the social classes in beginners in crime vanishes with age.
There are three to four times more lower-working-class offenders with a
debut in crime at age 13, whereas the number starting in crime at age 25 is
about the same in all social classes.
48 3. The Urban Offender
All in all, just as for the sex differences, the social class differences in
crime tend to narrow in many respects with increasing age of the cohort. It
might be possible to have a hypothesis about the reasons for this similar to
the one for sex differences. The protection from criminogenic influences in
youth might be class-related. Parental control and the kind and amount
of group activities might vary among the social classes, assuming, for in-
stance, that parental control is generally greater and group activities are
fewer and less delinquency-oriented in higher social classes. It might
also be the case that the social class of origin becomes less important with
increasing age.
All in all, it may be that differences in the rate of offending between the
two sexes and among different social classes, are partly related to general
variations among members of these groups in the degree of social control
they are subjected to during early youth, the most criminiogenic period in
life (LeBlanc & Frechette, 1989, p. 81), a period which may be critical to
the .development (or lack of development) of bonds to conventional society
(Hirschi, 1969).
4
The Social Contexts of Urban Crime
There are many ways in which crimes may be classified (Glaser, 1974,
pp. 45-78). The legal definition of crime is a common point of departure for
crime classifications. The legal definition specifies kinds of behavior that
are illegal (e.g., hitting, vandalizing, and stealing) and sometimes specific
under what conditions they are illegal but normally gives rather limited
information about the social contexts in which a criminal act takes place.!
The social circumstances of legally defined categories of crime can be very
heterogeneous. Therefore, a study of the social contexts of legally defined
categories of crime may in itself contribute to our understanding of the
behavior in question and its causes. Moreover, a crime classification based
on the social contexts of crime may be a more fruitful point of departure
than a legal classification for other types of analysis (e.g., ecological and
environmental analyses of the crime distribution and its correlates).
In this chapter, a classification and analysis of the social contexts of
selected traditional crimes will be made. The study covers violent crimes
(i.e., homicide, assault, violence to a public official, rape, robbery, and
molestation), vandalism, and serious property crimes (i. e., burglaries and
thefts of and from cars).2 The basic data analyzed are taken from the
Stockholm 1982 Crime Survey, a detailed study of a sample of about 7,500
police crime investigation records. The figures presented are weighted to
1-year figures by the different sampling fractions applied (see Appendix
A).
1. The same holds true, to a large extent, for the official crime statistics, which are
normally based on the legal definitions of crime.
2. The legal definitions of the studied crimes are given in Appendix B. Molestation
is not really a violent crime but rather a crime against the peace and will not be
specifically discussed. The social contexts of the crime of molestation have pre-
viously been studied by Nilsson (1988).
49
50 4. The Social Contexts of Urban Crime
Urban Violence
Crimes of violence occur in a variety of social contexts. Men assault their
wives at home, people are hit or robbed in the street by strangers, women
are raped by causal acquaintances, police officers are attacked in the course
of their duty, and so forth. Of the three major types of criminal behavior
dealt with in this study-violence, vandalism, and serious theft-the social
contexts of violence are likely to be the most heterogeneous ones.
Although violent crimes are heterogeneous, there seem to be some general
similarities in that much violence appears to be related to situations of
social control or friction between people.
First of all, criminological research shows that the overwhelming major-
ity of assaultive crimes develop out of a conflict situation. This is an experi-
ence found in Swedish studies (e.g., Wikstrom, 1980, pp. 125-127; Kuhl-
horn, 1984, p. 111), British studies (e.g., McClintock, 1963, p. 33), and
North American studies (e.g., Wolfgang, 1958, p. 191; Pittman & Handy,
1964, p. 467). Violence, especially assaults, may thus be viewed as one
way, among many others, to react to friction with other people.
Second, it seems that many violent actions might be interpreted as
aspects of social control (Black, 1984). Judging from the reading of state-
ments in police investigations, this is not only so in crimes of violence in
connection with police arrests, being refused entrance to a restaurant and
so forth, but also in many cases of violence within the family and even to
some extent in the violence that occurs in streets and other public places.
Violence is one means, among others, to try to influence the behavior of
others, or to punish others' undesirable behavior.
Intimate
relation
Family
violence
Public
place
__________-+__________ Residence
Street violence
attacks
Strangers
FIGURE 4.1. The private-public dimensions of crimes of violence.
description of the social contexts will be the scene of the crime and the
victim-offender relationship. There was no possibility to consider preced-
ing actions because no such data were collected for this study. However,
some knowledge of these points exists from previous Swedish studies made
by the author (Wikstrom, 1980, 1987c), and that information will be used
in the discussion of the various social contexts of violence. Before describ-
ing the contextual classification employed in this study, I shall discuss
McClintock's classification, because it is the major source of inspiration for
the one developed for this study.
McClintock's Classification
A contextual classification that partly builds upon the involved persons'
interpersonal relationship and the place of crime commission is that from
McClintock's (1963, pp. 29-30) London study of crimes of violence (with-
out any moment of theft). McClintock points to several of the empirical
problems of making a classification: the heterogeneous and complicated
nature of the violent event, the difficulties in getting precise information
concerning the circumstances, the discrepancies between the offenders'
and their victims' descriptions of these circumstances, and the problem of
getting information on the reason or the motive for the violence (pp. 27-
28). McClintock's classification is shown in Table 4.1.
A study of McClintock's classification shows that there appears to be
some overlap among the classes, a fact acknowledged by McClintock him-
self. He states that Classes 1 and 2 are motive-based and have priority over
the remainder, which are said to be based on circumstances. However,
even if one takes into account that Classes 1 and 2 are given priority over
the rest, it still seems possible to have some overlap among the classes.
Although perhaps not so common, the cases described in Class 3 may
occur both in public houses (Class 4) or at public places (Class 5). It is
Source: Crimes of Violenc.e (p. 33) by F.H. McClintock, 1963, London: Tavistock. Used by
permission.
Urban Violence 55
TABLE 4.2. McClintock and Gibson's classification of the social contexts of rob-
bery; main classes.
Class Description
Source: Robbery in London (pp. 14-15) by F.H. McClintock and E. Gibson, 1961, London:
Macmillan. Used by permission.
assumed here that Class 3 in fact have been given priority over Classes 4
and 5. Moreover, Class 2 may as equally well be based on circumstances as
on motives. It may finally be noted that Class 3 appears to use the victim-
offender interpersonal relationship as a major defining variable, whereas
Classes 4 and 5 obviously are defined according to place of occurence.
McClintock refers, in a later work (1974), to the classification as primarily
based "on the overall situation in which the incident occurred and on the
previous relationship of the parties involved" (p. 14). Although McClin-
tock's classification is the best that I have found in the literature, it is, as
has been demonstrated, not without problems. Therefore, for the present
study I have developed a modified version of McClintock's classification to
describe the social contexts of crimes of violence.
The classification by McClintock excluded crimes of violence with a mo-
ment of theft (Le., robberies). McClintock and Gibson (1961) had earlier
developed a separate classification for crimes of robbery (see Table 4.2).
However, because I wanted to have the same basic classification for all
kinds of violence, in order to be able to make direct comparisons of the
social circumstances among different kinds of violent behavior (Le.,
assault, rape, and robbery), the McClintock-Gibson classification for rob-
beries was not considered.
The second class refers to all cases of violence between nonfamily ac-
quaintances that takes place in residences. It should be pointed out that all
cases between nonfamily acquaintances occurring in the proximity of resi-
dences (i.e., cellars, staircases, or house lots) are also included in this
category.
The third class consists of all nonfamily cases of violence in places of
public entertainment (restaurants, bars, etc). Also included in this category
are cases that took place in the entrance area of the establishments, which
was quite a few of the cases in this category (see below). This class is
divided into three subcategories determined by the victim-offender inter-
personal relationships:
1. Relationships between nonfamily acquaintances.
2. Relationships involving a person (victim or offender) on duty or in the
course of work (i.e., police, doormen, restaurant personnel, etc.).
3. Relationships between strangers or causal acquaintances.
In referring cases to subclasses, the second category is given priority over
the rest, which almost always means over the third subclass, because nor-
mally the parties are strangers to each other in cases involving a person on
duty or in the course of work. The reason for specifying cases in which
persons are involved in violent encounters with others as part of their work
(e.g., police officers) or in the course of their work (e.g., a shop employee
who tries to intervene in a case of shoplifting) is that the social situation of
the crime can normally be regarded as very different from that when pri-
vate persons get involved in fights with causal acquaintances or strangers.
Causal acquaintances are classified as those with only a short duration of
acquaintance before the crime incidence (e.g., cases in which the victim
and offender met for the first time one evening in a restaurant and later on,
in that same evening, the victim was beaten up, robbed, or raped by the
offender).
The fourth class contains all nonfamily cases of violence in streets,
squares, and other public places. This class is divided into the same subcate-
gories as the third class, and the same principles to refer cases to subclasses
have been employed. All nonfamily cases occurring on public transport
and in its stations are included in this category.
The fifth, and final, main class contains all nonfamily cases of violence
under other circumstances than those covered by the four first classes.
It was divided into the same three subcategories, according to victim-
offender relationship, as Classes 3 and 4.
Violence Robbery
Social Serious Petty toward public against
context violence a assault official persons Rape Molestation Total
Within in family 18 21 0 1 10 7 13
Acquaintances
in residences 21 14 0 5 11 32 14
Places of public
entertainment 7 18 13 1 2 6 12
Acquaintances 2 0 0 1 2
Involving someone
on duty/at work 2 8 13 0 0 6
Strangers/casual
acquaintances 4 8 0 1 3 5
Public places 30 34 61 71 25 31 42
Acquaintances 2 4 3 3 2 6 4
Involving someone
on duty/at work 2 7 58 5 0 8 13
Strangers/casual
acquaintances 26 23 0 62 23 17 25
Other circumstances 24 13 25 22 52 24 19
Acquaintances 2 2 1 1 2 2
Involving someone
on duty/at work 5 4 24 0 2 4 6
Strangers/casual
acquaintances 17 7 0 21 29 18 11
assaultive crimes added together (i.e., serious assault, petty assault, and
violence toward a public officiaP) is shown in Table 4.4.
Around one fourth of all reported violence occurred in what may be
labeled as private (i.e., within the family or among acquaintances in apart-
ments), a bit more than one half occurred in public (i.e., in places of public
entertainment and other public places), and less than one fourth occurred
TABLE 4.4. Social contexts of assaultive crimes: (serious and petty assault and vio-
lence to a public official combined). Columne percent (Stockholm, 1982).
Assaultive crimes
Social context (N = 4,864; missing 261)
those with a previous family relationship (i.e., those who have split up)
may have a much lower dark figure than those having an ongoing relation.
In general, for closer acquaintances the dark figure is likely, because ofthe
less visibility, to be higher for cases occurring in residences than in other
places. The importance for the dark figure of the visability of the crime lies
in the differential likelihood of the presence of witnesses who can report
the incident to the police. A previous study by the author (Wikstrom, 1985,
p. 37) showed that it was the victim who notified the police in the majority
of cases of assault (67%), followed by a private person not on duty (18%).
In the same study, it was shown that the proportion of cases in which a
person other than the victim notified the police was highest for crimes tak-
ing place in streets and squares (40%) and lowest in crimes taking place in
apartments (28%).
The fact the the distribution of cases by social context is biased by the
differential selection of cases for reporting does not, however, necessarily
mean that the description of the modal circumstances of the various con-
texts will also be biased.
One obvious bias, though, is that the cases included, on average, will be
of a more serious nature than what holds true for all cases (i.e., reported
and nonreported). Because the crime circumstances and the characteristics
of the involved persons appear to show some difference between serious
and less serious cases of violence,4) and because the cases missed (i.e.,
nonreported cases) generally are of a less serious nature, one is to expect
some difference in the distribution of crime circumstances and involved
persons' characteristics between reported and nonreported cases.
4. The comparison (see Tables 4.7, 4.10, 4.13, and 4.16 below) of serious and petty
assault show in some instances significant differences. For example, the offenders in
cases of more serious violence tend more often to have an extensive record of
previous crime and violence than the offenders in cases of petty assault.
60 4. The Social Contexts of Urban Crime
good example of this perspective. Dobash and Dobash emphasize the im-
portance of the historically and socially constructed male dominance and
control for the understanding of wife assault (p. 15).
The social correlates perspective tries instead to explain, for example,
why some husbands, and parents are violent against their wives and chil-
dren and why others are not. A good example of research dominated by
this kind of question is that by Straus, Gelles, and Steinmetz (1980).
Within the role perspective, the importance of the legal and social
acceptance of violence in the family is often stressed, whereas the impor-
tance of factors causing friction between the family members plays a minor
role. Within the social correlates perspective, different conditions associ-
ated with the use of violence in the family are stressed. The various condi-
tions may be assumed to affect the violence within the family indirectly as
causes of friction between family members. The main ones pointed to in
the literature appear to be the following:
Low socioeconomic status.
Unemployment.
Pregnancy.
Single parenthood.
Alcohol and drug abuse.
Criminality.
To the extent that these aSSOCiatIOns have been studied in "normal"
populations, the correlations found are generally weak or moderate.
A special cause of the differential use of violence in the family that is
strongly pointed out by many researchers studying the social correlates of
family violence is the transmission of violence, through social learning,
from the family of origin to the present family (e.g., Straus et aI., 1980,
pp. 74, 100, and 122).
In the present study, a rather broad concept of the family is used, involv-
ing also those with a previous family relationship, if that relationship is not
too far back in time (i. e., a couple of years or so), and mothers- and
fathers-in-law. One half of the cases in this group occurred between mar-
ried or cohabiting persons, whereas formerly married or cohabiting per-
sons formed the next largest group (see Table 4.5). The overwhelming
majority of cases of violence within the family occurs in apartments (see
Table 4.6), and nearly one half occur in the home of both involved parties.
The next most common place of occurrence is the victim's home. There is a
clear relationship between the place of occurrence and the type of family
relationship in that the cases involving those with a previous relation occur
to a large extent in the home of the victim (normally the female). Females
dominate heavily among the victims of family violence. Most perpetrators
and victims are older than 30, and many are older than 40 (see Table 4.7).
The cases of violence among parents and children are few, but here we
have to expect a very high dark figure. In fact, in the present study, there
are more cases of children turning violent against their parents than the
Urban Violence 61
TABLE 4.5. Crimes of violence within the family by type of crimea and victim-
offender relationship (Stockholm, 1982). Percent.
Robbery
Serious Petty against
Victim-offender assault assault person b Rape Molestation All violence
relationship (n = 43) (n = 785) (n = 14) (n = 14) (n = 81) (n = 937)
TABLE 4.6. Crimes of violence within the familya by type of crimeb and place of
occurrence (Stockholm, 1982). Percent.
Robbery
Serious Petty against
Place assaultC assault person d Rape Molestation All violence
of occurrence (n = 43) (n = 785) (n = 14) (n = 14) (n = 81) (n = 937)
Other place of 16 4 0 0 11 10
occurrence
Sex
Both males 12 5 (25) 0 0 5
Male against female 60 92 (75) (100) 71 89
Female against male 28 2 0 0 25 5
Both females 0 0 0 4 1
Offender's age
19 and below 0 1 0 0 4 1
20-29 16 24 (33) (25) 22 23
30-39 44 39 (33) (50) 22 38
40 and above 40 36 (34) (25) 52 38
Victim's age
19 and below 0 4 (25) 0 0 4
20-29 16 21 (25) (46) 19 21
30-39 40 39 0 (39) 46 39
40 and above 44 36 (50) (15) 35 36
Nationality
Both Swedes 79 66 (67) (62) 58 66
Only offender
foreigner 4 9 0 (21) 11 9
Only victim
foreigner 0 6 (33) 0 23 7
Both foreigners 17 18 0 (17) 8 17
Previous criminality
Neither 36 31 0 (37) 64 34
Offender only 36 43 (67) (46) 27 41
Victim only 4 10 0 0 5 9
Both 23 16 (33) (17) 4 16
Extent of previous criminality
Offender
None 46 42 0 (39) 68 44
Recidivist 29 26 (67) (31) 16 26
Persistent 25 32 (33) (31) 16 31
Victim
None 74 75 (50) (81) 92 76
Recidivist 4 17 (50) (11) 8 16
Persistent 22 8 0 (8) 0 8
Previous violence
Offender
% with assaultb 29 34 (67) (27) 16 32
% with robbery 13 4 0 4 0 4
% with molestation 13 21 (33) (19) 12 20
Victim
% with assaultb 17 3 0 0 4 4
% with robbery 9 1 0 0 0 1
% with molestation 14 3 0 0 0 3
62
Urban Violence 63
family violence (see Table 4.7), cases between two foreigners are highly
overrepresented, given what can be expected according to their share of
the population. Judging from the reading of the statements in the police
record, this may to some degree be culturally related: Conflicts concerning
a woman's position in the family appear to be a rather common precedent
to the violence.
Although there are some, family violence cases reported to the police
from other than the lowest socioeconomic strata are unusual (Wikstrom,
1980). However, this holds true for all kinds of violence, although among
the victims of violence in public there tends to be a higher proportion of
middle-class people.
Perhaps of greater significance is the large involvement in family vio-
lence of individuals with a record of social problems of different kinds. For
instance, it is predominantly males with a previous criminal record who
are violent against their female family members. And many of the offen-
ders (around one third) have a previous record of violence (see again Table
4.7).5
No data about the relationship between violence and intoxication or
alcohol abuse was collected for this study. However, previous Swedish stu-
dies make it clear that a large proportion of the male offenders in family
violence are known for having alcohol problems. In one study (Wikstrom,
1980) it was found that 44 % of the offenders in family violence were known
for having alcohol problems, and as much as 23% had prior to the violence,
been treated at an institution for alcoholics. These are significantly higher
proportions than for all other contexts of violence, with one exception:
nonfamily violence between acquaintances in apartments.
Judging from the statements in the police investigations, conflicts con-
cerning drinking behaviour and jealousy seem to be common precedents to
family violence. It is often clear from the police material that the (in-
cluded) reported case is not the first one, but often has been preceded by a
number of incidents, many not reported to the police. In this connection it
might be of interest to mention that Dobash and Dobash (1984)-when
they compared police-reported cases of wife assault to descriptions from
interviews with assaulted wives of the first, the worst, and the last violent
incident-found that the pattern of police-reported cases corresponded
most closely to that of the last incident (p. 274).
As mentioned earlier, cases of parental violence against children re-
ported to the police are rather rare. Other Swedish studies of parental
violence toward children recorded by the police or the social welfare agen-
cies show, however, that this kind of violence, when serious, is generally
Robbery
Serious Petty against
assault assault persons Rape c Molestation All violence
Place of occurrence (n = 48) (n = 530) (n = 45) (n = 16) (n = 384) (n = 1,026)
Apartmentlproximityd 100 99 100 100 100 99
Home of both 4 14 0 3 9
Offender's home 7 10 0 3 8
Victim's home 68 57 77 94 71
Other's home 21 18 23 0 11
Robbery
Serious Petty against
Victim-offender assault assault person Rape Molestation All violence
relationship (n = 48) (n = 530) (n = 45) (n = 16) (n = 384) (N= 1,026)
Nonfamily
Relatives 7 8 0 0 2 5
Close acquaintances 89 71 54 91 24 54
Acquainted by
appearanceb 4 21 46 9 74 40
aSee notes to Table 4.8.
bFor molestation, most cases (84%) in the category "acquainted by appearance" refer to
cases between neighbors.
66 4. The Social Contexts of Urban Crime
Sex
Both males 86 49 77 0 33 47
Male against female 14 39 15 (100) 42 38
Female against male 0 6 8 0 11 7
Both females 0 7 0 0 14 8
Offender's age
19 and below 0 2 0 (3) 2 2
20-29 18 27 17 (19) 20 24
30-39 50 37 67 (64) 31 38
40 and above 32 34 16 (14) 47 36
Victim's age
19and below 0 8 0 (19) 4 6
20-29 21 16 0 (47) 21 18
30-39 36 36 54 (22) 16 29
40 and above 43 40 46 (12) 59 47
Nationality
Both Swedes 71 68 75 (31) 68 68
Only offender foreigner 7 13 8 (23) 11 12
Only victim foreigner 4 4 0 (11) 15 8
Both foreigners 18 13 17 (35) 6 12
Previous criminality
Neither 17 33 0 (43) 79 47
Offender only 13 24 50 (36) 20 24
Victim only 9 7 0 (11) 0 4
Both 61 35 50 (11) 1 25
Extent of prevo criminality
Offender
None 24 45 0 (52) 79 54
Recidivist 8 26 10 (24) 11 19
Persistent 68 29 90 (24) 10 28
Victim
None 31 65 50 (77) 96 74
Recidivist 19 23 17 (19) 4 15
Persistent 50 12 33 (3) 0 11
Previous violence
Offender
% with assaultb 48 34 90 (24) 14 30
% with robbery 28 2 20 (3) 0 4
% with molest. 36 19 50 (17) 8 17
Victim
% with assaultb 28 15 9 (3) 3 10
% with robbery 4 0 0 (0) 0 0
% with molest. 12 7 9 (3) 1 5
ment activities. The experience from the present study shows that most of
the violence in public not occurring in places of public entertainment or
their entrance areas occurs while people are on their way to or from public
entertainment activities or in areas with many places of public entertain-
ment. The spatial and temporal pattern of assault in public is also fully
consistent with its relation to public entertainment activities-that is, most
occurs in areas with many public establishments and in the late evening or
at night. Finally, the reading of the various statements in the police inves-
tigation confirmed that much of the violence in public is related in one way
or another to public entertainment activities.
Several British victim surveys show that the risk of violence is closely
related to participation in public entertainment activities. Sparks, Genn,
and Dodd (1977, p. 105) report that of those interviewed who were never
out in the evenings, 2.3% had been subjected to violence, whereas the
corresponding figure for those out every evening was 22.2%. Similar
results have been reported in newer British study by Gotffredson (1984,
p.lO).
For the reasons mentioned above, it was decided to discuss violence in
places of public entertainment jointly with violence in other public places.
First, I shall give some general comments on violence in public and there-
after discuss somewhat more in detail assaults, robberies, and rapes in
public.
About one fifth of all violence in public occurs in places of public enter-
tainment, and about one fourth occurs on public transport. The violence
on public transport predominantly takes place in the underground and its
stations. Less than one half occurs in streets, squares, and parks (see Table
4.11). Regarding different types of violent behaviors and comparing assaul-
tive crimes with rapes and robberies, a major difference is that the latter
are strongly concentrated in streets, squares, and parks (around 80%),
whereas only around one third of the assaultive crimes (considered jointly)
occur in streets, squares, and parks.
Most violence in public occurs between strangers (see Table 4.12).
Moreover, most violence between strangers occurs in public in the course
of public entertainment, a fact that also means that strangers rather seldom
get involved in violence outside public entertainment (see again Table 4.3).
A high proportion of the violence in public involves someone on duty or
in the course of work, as either victim or offender. Of all reported cases of
violence in public, as much as 27% involved someone during work, in 55%
of these cases as victim and in 45% as offender. However, one must allow
for the fact that the reportability of those victimized while working may
be higher than for those victimized as private persons. The five groups
with the highest number victimized during work was, in order of rank,
doormenlwatchguards (predominantly doormen), police officers, shop em-
ployees, public transport employees, and restaurant personnel (excluding
doormen). This finding is in line with what has been reported from the
TABLE 4.11. Crimes of violence in public places by type of crime and place of occurrence (Stockholm, 1982). Percent.
f'-
;J
o
t;/)
o
!:l.
a
()
::l
...oo
TABLE 4.12. Crimes of violence in public places by type of crime and victim-offender relationship (Stockholm, 1982). Percent. ~
'J>
Swedish census bureau's victim surveys (SCB, 1981). The five groups with
the highest number of individuals offending in the course of their work
was, in order of rank, doormenlwatchguards, restaurant personnel, public
transport employees, police officers, and shop employees. These figures
illustrate that much of the violence in public is linked to social control
situations. Comparing assaultive crimes (serious and petty assaults and vio-
lence toward a public official) with robberies and rapes, as might be ex-
pected, almost all cases involving a person on duty or at work belonged to
the group of assaultive crimes, meaning that the proportion of the assaul-
tive crimes involving someone on duty or at work is significantly higher
than for all violent crimes taken together.
f"
....,
::r
(1)
TABLE 4.13. Crimes of violence in public places by type of crime and victim and offender characteristics (Stockholm, 1982). a Percent. t:Il
0
("l
~
CJ>
Sex
0
....,
Both males 91 80 78 89 0 56 78
Male against female 6 16 9 8 100 38 16 c::
....
0-
Female against male 2 2 8 2 0 3 3 ~
0
Both females 0 3 5 0 0 3 3 n
::1.
Offender's age 3(1)
19 and below 11 14 10 39 24 12 14
20-29 48 39 47 35 24 35 40
30-39 26 25 27 24 28 32 27
40 and above 15 22 16 2 24 21 19
Victim's age
19 and below 15 15 0 8 13 16 11
20-29 44 41 75 29 45 31 43
30-39 27 20 18 22 13 26 21
40 and above 14 24 7 41 29 27 25
Nationality
Both Swedes 50 60 76 67 37 68 66
Only offender foreigner 29 22 22 14 37 10 20
Only victim foreigner 0 10 2 7 5 11 7
Both foreigners 21 8 0 12 21 11 7
Previous criminality
Neither 18 24 32 14 35 56 28
Offender only 27 34 63 43 50 32 42
Victim only 14 17 1 12 0 5 11
Both 41 25 3 31 15 7 19
Extent of previous criminality
Offender
None 28 42 32 34 35 58 40
Recidivist 32 33 33 25 45 15 30
Persistent 40 25 34 41 20 26 30
Victim
None 46 62 95 55 93 88 69
Recidivist 28 27 4 27 3 8 21
Persistent 26 11 1 17 4 4 9
Previous violence
Offender
% with assault b 52 32 34 31 40 29 33
% with robbery 12 5 5 11 5 7 6
% with molestation 32 14 20 32 15 20 19
Victim
% with assault b 37 18 2 11 6 5 13
% with robbery 9 1 0 3 0 2 1
% with molestation 14 9 8 3 6 c:::
....
0-
Il>
-See notes to Table 4.8. ::I
b Including few, if any, cases of homicide and rape.
~
0
(b
::I
("l
(1l
-..l
W
74 4. The Social Contexts of Urban Crime
1978, p. 326). As already noted above, a high proportion of the cases in-
volve socially marginalized persons, and some research indicates that peo-
ple who become aggressive after drinking alcohol are especially found
among individuals with social problems, which is often related to low self-
esteem (e.g., Boyatzis, 1975, p. 1205; Deardorff, Melges, Hout, & Sava-
ger, 1975, pp. 1193-1194).
In cases of assault in the public by youth, it is not uncommon that the
violence is preceded by molestation (e.g., forcefully "begging" for
cigarettes). Sudden attacks are comparatively rare, but when they do occur
most of the offenders are youth (Wikstrom, 1980).
Most of the assaults in public occur between people with different back-
grounds. For instance, around one half of the assaults occur between one
individual with a previous record and one who lacks such record, and
almost one third of the cases are between a Swede and a person of foreign
nationality (see again Table 4.13).
In connection with this, it might be mentioned that a fairly recent Swed-
ish study of criminals in prison showed that, while free, the individuals
were likely to be frequent participants in public entertainment (Aker-
strom, 1983). Moreover, Swedish study of the relationship between leisure
activities and criminality among youth in Stockholm shows that the most
criminal of the youths tend to engage more in commercial spare-time
activities and spend more time in the inner-city areas than other youth
(Sarnecki, 1983).
Against the background of what has been discussed, there is much that
points to the fact that a major reason for the comparatively high rate of
assaults in the course of public entertainment is that, when a great number
of strangers with different backgrounds are brought together in a small
area and most of them are intoxicated, there is a high rate of friction that
sometimes develops in to violent actions. Moreover, public entertainment
is also likely to produce a high number of social control situations, which at
times may develop into violence.
As a final comment concerning assaults in public places, it may be men-
tioned that although most of this violence is related to public entertain-
ment, there is a great difference between crimes occurring in the day and
those occurring at night. Assaults in public during the day are rather infre-
quent, but when they occur they are mostly socially marginalized persons'
internal confrontations or their confrontations with different kinds of
authorities, shop employees, and "conventional" people.
Judging from the case descriptions and statements in the police inves-
tigations of these cases, as a rule, the most planned robberies are those in
which the offender, or offenders, decide to rob someone and wait for a
suitable target to show up, or move about to find a good place and target to
rob. In other, less planned cases, there appears more to be a situational
inducement to rob, that is, a good opportunity suddenly arises, such as
when an older alcohol abuser passes a group of youth sitting on a park
bench. A third type of robbery appears to have no preplanning at all.
Rather, a fight may end with the winner taking a wallet or other valuables
from his antagonist after knocking him down. It is not uncommon that a
case that first appears as a robbery in the police report may be redefined as
an assault and a theft when it appears in court (Andersson, 1984, p. 213).
In contrast to assault, robbery is by definition a crime of gain, although a
question mark may very well be put as to whether the theft really is the
main reason for the crime. Pratt (1980) states, in a London study of
mugging,6 that "the actual financial gain from mugging is normally so small
that it must be seen as totally incidental to the main object of the opera-
tion, which, I would suggest, must therefore be to demonstrate toughness
and masculinity" (pp. 164-165). Pratt also claims that a street robbery
(i.e., a mugging) is "not so much a criminal activity as an attempt to estab-
lish a social identity, and in this sense I would argue that mugging bears
close resemblance to the mods/rockers/hippies/skinheads and other similar
phenomena" (p. 62). In this connection, it may be of interest to mention
that one third of the offenders below the age of 25 in street robberies in
Stockholm were noted in the investigations as belonging to different kinds
of youth groups, such as skinheads or rockers. Moreover, in the majority
of the street robbery cases, the value of the stolen goods was less than 500
SEK (approximately 45 or $75).
However, the reasons for committing a street robbery may differ among
different age groups. In a U.S. interview study of 113 robbers from Califor-
nia, Feeney (1986, p. 56) reports that although the majority of the adult
robbers (64%) stated gain as major motive for the robbery, that was so
only for a minor part of the young robbers (35%).
Pratt's (1980) discussion of London muggings appears mainly to be ap-
plicable to the street robberies committed by youth. A large part of the
street robberies in Stockholm are committed by youth; 4 out of 10 offen-
ders are below the age of 20, and more than one half are below the age of
25. Almost one third of the street robberies are committed by groups of
youth. There are good reasons to believe that the actual proportion of
youth and youth groups in street robberies is higher than what one finds
in a study of reported crimes. (For a discussion, see Wikstrom, 1987b,
Sections 4.2.3 and 4.2.4.)
6. Pratt (1980, p. 38) defines mugging as robberies in public places between stran-
gers.
76 4. The Social Contexts of Urban Crime
Judging from the case descriptions and statements in the police investi-
gations, most rapes in public appear to involve some kind of planning,
although many times it appears to be a minimum of planning. This finding
is, by and large, consistent with that of Amir's (1971, pp. 141-143)
Philadelphia study of rape, in which he found that most rapes involved
some kind of planning.
Most of the rapes in public are attempts (73%) and some of these have
more of the character of a sexual molestation than a rape attempt. These
findings are in close correspondence with previous Swedish studies of rape
(see Persson, 1981, pp. 63-64). In most cases (82%), the duration of the
event is less than 5 minutes from the point in time when the victim and
offender meet until it is over. This also means that most rapes in public are
sudden attacks. The generally short time span of the act is related to the
fact mentioned earlier that most rapes in public are attempts.
One half of the rapes in public occur in streets and squares, one third
occur in parks, and the remaining are spread out among various places.
Cases in streets, squares, and parks almost exclusively involve strangers,
whereas cases in places of public entertainment, shops, or other places of
service mostly involve parties who are acquaintances (e.g., workmates).
Compared to assaults and robberies in public, the offenders in rape cases
are less concentrated to certain age groups, unlike their victims, who show
a concentration to the 20s. A very high proportion of the cases involve a
foreign offender. In fact, the great majority of the offenders are foreigners.
Also, the proportion of cases occurring between two foreigners is high in
regard to what can be expected given their proportion of the population.
The offenders are to a high degree previously recorded for crime, and a
very high proportion are previously recorded for violence. However, the
victims of rape in public are generally unrecorded (see again Table 4.13).
The reason for the fact that sometimes in the tables of victim and offender
characteristics, the proportion of cases in which both victim and offender
Urban Violence 77
are previously recorded are higher than the number of cases where the
victim is previously recorded is that the victims to a higher degree are pre-
viously recorded in cases with a known offender than in cases where no
offender has been apprehended. In this connection it may be mentioned
that the actual number of studied cases of rape is twice as high as the
numbers given in the tables due to the weighting to I-year figures (i.e., all
cases of rape during 2 years were included-see Appendix A).
7. Although in this group of crimes it was sometimes difficult to establish the de-
gree of relationship with any greater certainty. It is likely that the proportion of
strangers in this group of crimes is overestimated.
-J
00
f'-
;l
(t>
CIl
o
(')
.:
g
::s
8"
TABLE 4.14. Crimes of violence in other circumstances by type of crime and victim-offender relationship (Stockholm, 1982). ~
o
....,
% Violence % Robbery C
% Serious % Petty toward against ....
0-
po
Victim-offender assault assault public official person % Rape % Molestation % All violence ::s
relationship (n = 56) (n = 485) (n = 219) (n = 206) (n = 74) (n = 288) (N=1,328) n
~.
Close acquaintances 3 12 0 3 6
3
(t>
5
Acquainted by appearance 6 5 8 0 11 7
Casual acquaintances 21 8 0 17 30 2 9
Strangers 70 75 92 80 68 82 78
TABLE 4.15. Crimes of violence in other circumstances by type of crime and place of occurrence (Stockholm, 1982).
% Violence % Robbery
% Serious % Petty toward against
Place assault a assault public official persons % Rape % Molestation % All violence
of occurrence (n = 56) (n = 485) (n = 219) (n = 206) (n = 74) (n = 288) (N= 1,328)
Apartment/proximity b 60 39 13 62 59 57 44
Home of both 6 3 1 0 0 0 1
Offender's home 6 6 5 0 19 5 6
Victim's home 33 18 0 42 30 47 26
Other's home 15 12 7 20 10 5 11
Wood or greater park area C 0 3 0 5 14 0 3
Public institution 27 27 37 4 3 4 19
School/school yard 0 6 0 2 2 1 3
Hospital/medical care centre 21 7 4 2 0 1 5
Other medical institution 0 0 1 0 0 1 0
Police station 0 9 29 0 0 0 8
Other public institution 6 5 3 0 3
Other places 13 31 50 29 24 31 34
Hotel 0 6 0 5 7 4 4
Factory 0 0 0 0 0 0 0
Office 0 0 1 2 1 C
..,
cr"
Other workplace 0 3 0 0 0 1 P>
::I
Harbor or on board ship 3 5 3 3 1 4 4
Inside vehicle 0 9 36 3 12 21 15 $
0
Other places 10 8 10 16 3 8 9 (b
::I
('l
(1l
aHomicide and aggravated assault.
b Proximity = staircase, cellar, or house lot.
-.)
cOutside densely built-up area. Cases in parks within the densely built-up areas appear among the cases in public places context. \Q
~
~
~
I/)
TABLE 4.16. Crimes of violence in other circumstances by type of crime and victim and offender characteristics (Stockholm, 1982).- 0
O.
% Violence % Robbery e:.
Victim and offender % Serious % Petty toward against n
0
-characteristics assault assault public official person % Rape % Molestation % All violence ::s
....
~
><I
....
til
Sex
0
....,
Both males 75 74 65 80 0 53 68
Male against female 12 17 8 19 100 31 20 C
...,
cr"
Female against male 9 4 19 2 0 4 7 po
::s
Both females 3 5 7 0 0 12 5
n
::I.
Offender's age 3
~
19 and below 0 14 9 23 6 6 11
20-29 41 40 45 50 38 35 41
30-39 48 33 38 23 35 21 32
40 and above 11 13 8 4 21 38 16
Victim's age
19 and below 0 11 0 7 19 7 8
20-29 18 31 83 14 52 30 37
30-39 24 26 10 20 21 30 23
40 and above 58 32 7 59 8 33 32
Nationality
Both Swedes 38 63 75 68 39 64 64
Only offender foreigner 42 10 25 23 32 21 20
Only victim foreigner 4 10 0 0 12 13 7
Both foreigners 15 16 0 9 17 2 9
Previous criminality
Neither 30 30 27 16 22 62 34
Offender only 30 30 71 58 58 33 46
Victim only 9 25 2 10 9 2 10
Both 30 15 0 16 11 2 9
Extent of previous criminality
Offender
None 32 55 27 25 32 65 44
Recidivist 32 20 37 20 36 20 26
Persistent 36 25 36 55 32 14 30
Victim
None 59 53 98 69 85 94 74
Recidivist 15 32 2 12 12 4 16
Persistent 26 15 0 19 3 3 10
Previous violence
Offender
% with assaultb 50 24 44 50 43 12 32
% with robbery 7 4 4 35 11 0 7
% with molestation 32 8 27 50 22 12 20
Victim
% with assaultb 19 20 0 15 1 2 11
% with robbery 11 1 0 2 0 0 1
% with molestation 11 10 0 10 0 0 6
C
...
cr
aSee notes to Table 4.8. ~
lence toward a public official, took place inside vehicles, normally during
transport of a suspect after a police intervention.
Urban Vandalism
Unlike assaultive crimes, robberies, and rapes, vandalism is not a crime
against the person. Moreover, it is different from serious property crime
and robbery because it is normally not a crime of gain, although, as dis-
cussed in the previous section on urban violence, the gain motive appears
not always to be the main reason for the crime in robberies against the per-
son. Vandalism is a crime of destruction of property. Some authors have
used the term vandalism to cover all kinds of property destruction, both
legal and illegal (e.g., Roos, 1986, pp. 25-26). It may, however, be helpful
to restrict the word vandalism to illegal destruction of property, because
normally the social circumstances of legal and illegal destruction of proper-
ty are likely to be very different. The present study is concerned only with
illegal destruction of property, and the term vandalism will be used in that
sense.
Acts of vandalism cover a wide range of behaviors, for instance,
scratching the paint off cars with a key while passing by, covering walls or
underground train wagons with one's "tags," lighting fire in a cellar, and
damaging street lights. Also, the background to acts of vandalism varies
significantly. Some acts of vandalism may foremost be regarded as a con-
sequence or a part of committing another crime. The committing of a bur-
glary often requires some damage of property. In the course of an assault,
property may more or less accidentally get damaged, or an assault may be
combined with the destruction of the antagonist's property. Vandalism in
private often occurs in conjunction with other types of crimes.
As to the pure vandalism cases, the greatest part take place in public
places. Predominantly, this kind of vandalism can be described as
"weekend vandalism," a concept used by Roos (1986). Roos describes a
typical case in the following way:
This type of vandalism often consists of youth, but also older persons, on a
weekend evening or night, moving through the city on their way to areas of public
entertainment after having been at a private party, and then, for instance, kicking
against parked cars, throwing sausages and mashed potatoes in the face of a kiosk
employee so her glasses break, kicking in doors to dance halls to which they are
refused entrance, etc. (p. 119, my translation).
Source: Faktisk Brottslighet Bland Skolbarn (p. 132) by K. Elmhorn, Sou 1969:1, Vad Var Det
Vi Sa! (p. 91) by B. Olofson, 1971 Utbilonincsforlago, Stockholm. Used by permission.
"The study included 950 schoolchildren in Stockholm.
bThe study included 519 schoolchildren (boys only) in Orebro (a medium-sized Swedish city).
In apartments/proximity 15 35
In apartment 4
In staircase, attic, cellar, or house lot 11
In streets and squares 64 11
Against cars 24
Against buildings 35
Other 5
In public transport 11 1
Underground, railway (on board
or in station) 9
Other public transport" 2
Public locales 7 25
Public entertainment 1
School, sports arena, church, etc. 3
Shop or service station 2
Workplaces (other than above)
Other places 3 19
Total 100
Vandalism in Private
It is unusual that youths are involved in acts of vandalism in private, and
when they are, it is normally in damages made in cellars or staircases. The
offenders in vandalism in private are to a very high degree previously re-
corded for crime, and a rather large proportion-more than one tenth-
are known to be intravenous drug abusers. About one half of the offenders
are unemployed. This picture does not change very much whether one is
considering all cases or only cases of pure vandalism (see again Table
4.18).
The social circumstances of vandalism in private often resemble those
described earlier for assault and robberies in private. It is not uncommon
that these acts of vandalism occur during heavy drinking parties in connec-
tion with a fight or when some of the participants get a "delirium." Other
circumstances of vandalism in this context have to do with the parties' pre-
vious relationships; for instance, a former husband may be refused en-
trance to his former wife's apartment, and the damages emerge during the
husband's more or less successful attempts to enter the apartment, or, as it
looks, the damages may result from the husband's more aggressive ways of
showing dissatisfaction (destroying property) at not being let in.
Urban Vandalism 87
Offender's age
19 and below 4 27 16 9 30 22
20-29 42 33 36 45 33 44
30-39 36 28 16 27 26 22
40 and above 18 12 32 19 11 12
Offender's sex
Male 86 91 93 83 92 91
Female 14 9 7 17 8 9
Offender's nationality
Swedish 78 80 72 90 84 81
Foreigner 22 20 28 10 16 19
Offender extent of
previous criminality
None 19 46 40 14 45 38
Recidivist 34 30 32 33 29 38
Persistent 46 24 28 53 26 24
Previous vandalism 45 25 32 57 23 29
Offender known
intravenous drug abuser 14 4 19 14 4 18
Offender unemployed 46 28 44 53 33 36
Offender in youth group 2 19 4 4 21 5
aExcluding cases of vandalism in connection with other crime.
The two major reasons for committing thefts pointed to in the literature
are gain, and thrill and excitement. Gain is an obvious reason for commit-
ting crimes of theft, although for some types of crimes of theft (e.g., sho-
plifting), gain may not always be the major reason for the crime-thrill and
excitement may be just as important. However, of the more serious types
of crimes of theft included in this study, it is likely that gain, in most cases,
is an important reason for committing the crime. In an interview study with
150 Swedish inmates, focusing on thieves, Akerstrom (1983) sums up her
findings, as regards the reason for living a criminal life-style, in the follow-
ing manner:
At least in a limited time perspective, crime can be lucrative. Another rather evi-
dent facet of criminality is that it can be exciting, adventurous and filled with thrills.
Since criminals can be described as action-seekers, this is obviously a rewarding
factor. A third characteristic of a criminal lifestyle is that it is secret, which in itself
gives a certain amount of fascination. Finally, independence is discussed since this
so often was stated as important by the interviewees and thus given a value of its
own. (p. 221)
Some research findings suggest that the main motive for theft may differ
among age groups; that is, motives have more to do with thrill and excite-
ment at younger ages, and more with gain at older ages. For instance,
Belson (1975, pp. 145-170), in a study of 13- to 16-year-old London boys,
reports support for the idea that the seeking of fun and excitement plays a
significant role in juvenile theft. Belson states that "'fun and excitment
seeking' appears to be an important contributing factor operating within a
complex of other factors" (p. 169). Farrington (1986, pp. 231-232) reports
findings from several studies indicating that as subjects grow older, the
dominant reason for committing crimes of theft appears to change from
thrill to economic gain. Interviewing a sample of delinquents in adoles-
cence and early adulthood from Montreal, Canada, LeBlanc and Frechette
(1989, Chap. 2) show that although both "hedonistic" and "utilitarian"
motives for committing crimes of theft are common at all ages, the impor-
tance of gain increases with age, whereas the importance of thrill and
excitement decreases.
A special aspect of theft is that of theft as a means of financing an abuse
of drugs. As was presented in Chapter 3, a high proportion of the offenders
in crimes of serious theft are likely to be drug abusers, especially those
involved in residential burglaries and thefts of and from cars, and much
more so than the offenders in violent crimes and vandalism, with the excep-
tion of the robberies, which also had a rather high proportion offenders
likely to be drug abusers. All in all, it appears that drug abusers are more
represented in serious crimes of gain than in crimes of vandalism and vio-
lence, but also that there may be significant variation among types of
crimes of gain in the proportion of drug abusers involved.
The question of whether drug abuse causes a person to be a thief or
90 4. The Social Contexts of Urban Crime
whether a drug abuse follows upon the involvement in crime (for instance,
as a consequence of exposure to certain milieus and life-styles) has been
intensely debated. Regardless of the answer to that question, it is likely
that drug abuse accelerates the motivation to theft, due to the high costs of
financing drug addiction. Theft and drug-dealing appear to be the main
sources of finance, at least for male abusers. (For discussions, see Knutsson
& Kuhlhorn, 1979, pp. 16-20; Torstensson, 1987, pp. 146-148).
Place of theft
Home 1 3
Other place 2 4
Burglaries
The burglaries have been divided according to the major category of place
burglarized (see Table 4.19). Because the official crime statistics show
burglaries according to place burglarized, there was an opportunity to com-
pare the distribution from the sample of crimes included in the Stockholm
Residence/proximity 51 52
Apartment 26 25
Cellar/attic 25 27
Public locale 22 19
School, sport, etc. 5 5
Public entertainment 4 3
Shop, service station 13 11
Semipublic locales, etc. 20 23
Office/factory 10 10
Building site/garage 5 5
Kiosk/vending machine 5 8
Other places 7 6
a Excluding a smaller number of burglaries of safes because no place of occurrence for these
cases is given in the official crime statistics.
92 4. The Social Contexts of Urban Crime
1982 Crime Survey (which has been weighted to I-year figures) and the
official statistics including all crimes. The comparison shows a very high
degree of correspondence. About one fourth of the burglaries occur in
apartments, about one fourth occur in cellars or attics, about one fifth
occur in public locales, and about one fifth occur in semipublic locales (in-
cluding building sites and vending machines). The division between public
and semipublic locales is made on the grounds that the former are normally
open to the public, whereas the latter are generally only open to those who
work in the place or have an errand to the place.
Most of the residential burglaries were directed toward fiats in multistory
houses (75%), but if this figure is compared with the proportion of all
residences in Stockholm that are fiats in multistory houses (89%), this
finding indicates a somewhat lesser risk of burglarization for apartments in
multistory houses than for single-family houses.
In about two thirds of the residential burglaries (68%), the entry was
made through a door, and in the remaining one third (32%) the entry was
made through a window. Comparing burglaries in multistory houses and
single-family houses, the way of entering in the former is normally through
the door (84%), whereas entry through a window dominates (68%) in the
latter case.
Of the nonresidential burglaries, the way of entrance in nearly two thirds
(63%) was through a door, and in nearly one third (30%) through a win-
dow. In the remaining cases, the offender entered by the rough or through
a wall, and in the smaller number of burglaries of vending machines, the
way of entrance was, of course, irrelevant.
Comparing the offender characteristics by major type of place burgla-
rized (see Table 4.20) shows some interesting relationships of the offen-
der's age and pevious criminal involvement to the degree of privacy of the
place burglarized: 9
e The more private the place burglarized is, the older the offender tends to
be.
e The more private the place burglarized is, the more previously recorded
for crime, and the more previously recorded for crimes of theft, the
offender tends to be.
All in all, it looks like it takes a more experienced offender to burglarize
a private rather than a nonprivate space. Generally, the average gain is
also higher in burglaries of private rather than nonprivate spaces (see
Table 4.21), although the variablity among the cases is great. Table 4.20
also tells us that there is seldom big economic values that are lost in a single
burglary. However, there are a small number of cases with high gain (i.e.
over 40,000 SEKJ6,500 U.S. Dollars; normally these belong to the nonre-
Sex
Male 93 91 97 96 92
Female 7 9 3 4 8
Age
19 and below 2 7 27 29 15
20-29 58 50 43 43 39
30-39 27 33 20 23 36
40 and above 13 10 10 5 10
Nationality
Foreigners 9 12 21 12 15
Previous criminality
None 6 17 25 32 31
Recidivist 14 21 27 28 18
Persistent 80 62 48 40 51
Previous theft
% with previous theft 89 73 67 63 60
-School, sport place, locale of public entertainment, shop, locale of service.
bOffice, factory, building site, garage, kiosk, vending machine.
TABLE 4.21. Value of stolen cash and goods in residential and nonresidential burg-
laries. Columne percent (Stockholm, 1982).
Residential Nonresidential
0 72 30 87 29
1-500 13 5 7 13
501-2,500 10 15 4 38
2,501-5,000 3 20 10
5,001 and above 2 30 10
Note. 100 SEK is currently about 10 or $ 15.
sidential burglaries. From the point of view of the offender, the value of
the stolen goods is often much less because the "sales value" of stolen
goods is normally much below its actual value (Knutsson, 1980, p. 357).
Not retained 5
Retained, no damages 17
Retained, minor damages 32
Retained, big damages 12
Attempts 34
theft from cars were reported to the police. However, in cases of attempts,
it may be difficult to correctly classify the cases as either an attempted car
theft or an attempted theft from a car without any intention to steal the car
itself.
McCaghy, GIordano, and Henson (1977, pp. 378-380) have suggested
the following typology for auto theft:
1. Joyriding.
2. Short-term transportation.
3. Long-term transportation.
4. Profit.
5. Commission of another crime.
Classes 1, 2, and 5 refer to cases in which the offender did not intend to
keep the car, whereas Classes 3 and 4 refer to cases in which the offender
either intended to keep the car for himself or to sell it. The findings from
the study of auto theft in Toledo, Ohio, by McCaghy et al. (1977, p. 381)
showed that most (around 60%) of the car thefts were for joyriding or
short-term transportation. In LeBlanc and Frechette's (1989, Chap. 2)
study, of a sample delinquents from Montreal during adolescence and early
adulthood, it is reported that the motives stated for motor vehicle theft at
all studied ages are predominantly hedonistic, although there was a very
slight decrease in hedonistic motives as subjects grow older. No informa-
tion about the offender's motives behind stealing the car was collected for
the present study, although it was found that only a minor proportion of
the cars were lost (see Table 4.22)-most had been found in time for the
coding of the crime files. The coding was made, on average, about 1 year
after the time of crime commission. However, most cars were found in a
rather short time span after the crime commission.
The overwhelming majority of cars stolen (or attempted to be stolen)
were parked in public spaces, in particular on the street (see Table 4.23).
Moreover, the cars owned by private persons residing in Stockholm were
predominantly (66%) parked within 500 meters from the owner's residence
at the time they were stolen or an attempt to steal the car was made.
Also, in cases of thefts from parked cars, the most common place on the
Urban Serious Theft 95
TABLE 4.23. Place of crime commission: thefts of and thefts from cars. Columne
percent (Stockholm, 1982).
Place of crime Thefts of cars Thefts from cars
commission (n = 4,065; missing 130) (n = 16,797; missing 495)
TABLE 4.24. Items stolen in cases of thefts from cars (Stockholm, 1982).
Number
Item stolen (n = 14,718; missing 2,574) %
TABLE 4.25. Offender characteristics in car thefts and thefts from cars (Stockholm,
1982). Percent.
Offender characteristic Thefts of cars Thefts from cars
Sex
Male 95 95
Female 5 5
Age
19 and below 20 12
20-29 44 44
30-39 29 37
40 and above 7 7
Nationality
% foreigners 14 13
Previous criminality
None 18 33
Recidivist 19 24
Persistent 63 43
Previous theft
% with previous theft 69 62
street (see again Table 4.23), and most (62%) of the thefts from cars own-
ed by private persons residing in Stockholm occurred within 500 meters of
the owner's residence.
Most cars are likely to be parked on the street, and as long one does not
have information about the distribution of the frequency of where cars are
parked by type of parking place, the results shown in Table 4.23 cannot be
used to say anything about the relative risk of thefts of and from cars by
place of car parking.
The most frequently stolen objects in thefts from cars were car radios or
car stereos (see Table 4.24). In one half (52%) of all thefts from cars,
including those with nothing stolen, a car radio or stereo was stolen.
However, of all objects stolen, car radios and stereos total 45%. The
second most frequent stolen objects were parts attached to the car (e.g.,
instruments); the third, tools of different kinds. All in all, most things
stolen from cars were car items. In only 2% of the cases was the value of
the stolen goods above 5,000 SEKl800 U.S. Dollars.
As previously mentioned, in most thefts of cars the car was retained.
However, in around one third (38%) of these cases something had been
stolen from the car. Again, car radios and steroes were the most frequent
objects stolen, followed by parts attached to the car and tools.
The offender characteristics for car thefts as compared with thefts from
cars are very much the same, although the offenders in car thefts tend to be
a bit younger and more previously recorded for crime (see Table 4.25).
Conclusion 97
Conclusion
of the crime, more so for less serious than serious criminality. Generally,
crimes in private are likely to have a high dark figure and a relatively high
clearance rate, whereas crimes in public are likely to have a lesser dark
figure and a much lesser clearance rate. However, for crimes of theft of a
more serious nature, the dark figure is likely to be low and the clearance
rate very low, regardless of whether the theft is directed toward the private
or the public. As for crime prevention, suitable and effective measures are
likely to vary with the public-private dimension of crime. The recent em-
phasis on the value of situational crime prevention is one expression of
the importance that has been put on the social contexts of crime for find-
ing suitable and effective crime prevention measures (see, e.g., Clarke &
Mayhew, 1980), although situational crime prevention has predominantly
been concerned with crime in public.
The private and public sides of people's lives are often very different.
Expressions such as "behind closed doors" or "would never do that in
public" indicate the importance put on the division between private and
public in everyday life. It is therefore no great surprise that the social con-
texts of crime, and victim and offender characteristics, show significant dif-
ferences between the private and public, although, in this chapter I have
only been able to touch upon parts of the many relevant factors that are
likely to vary with the social contexts of the crime.
It will be suggested here that the private-public dimension of crime is
important for the understanding and prevention of crime and that a divi-
sion of traditional crime, building upon this dimension, is likely to turn out
to be a much more fruitful point of departure for analysis than a division of
crimes by legal definitions, for example, in the study of criminal careers or
ecological and environmental aspects on crime. Unfortunately, it was not
possible to employ this approach in the analysis of urban criminal careers
presented in Chapter 3. However, in the analysis of area crime patterns in
Chapter 8, a division of the crimes by their social contexts, focusing on the
private-public dimension, will be an important point of departure.
5
The Urban Victim
100
The Victimological Approach 101
tion that belongs to the social ecology of crime (see Chapter 8). Modern
crime ecology focuses on what brings offenders and targets together in a
crime situation (Brantingham & Brantingham, 1981, p. 8).
Against this background, it is hardly surprising that Steinmetz (1981)
finds Cohen and Felson's (1979) routine activity approach (discussed in
Chap. 2), which is essentially situational in nature, a useful point of depar-
ture for the creation of a model of personal victimization and prevention of
personal crimes.
Steinmetz (1981) uses concepts similar to those of Cohen and Felson
(1979)-potential offenders and potential victims-but instead of speaking
of capable guardians, he prefers the wider concept of barriers between the
victim and the offender. Capable guardians (i.e., formal and informal sur-
veillance by people) may in this context be regarded as one among many
different types of social barriers. Steinmetz distinguishes between social
and technical barriers and barriers through environmental design.
Using the language of victimology, Steinmetz (1981) differentiates
among three main factors influencing the risk of victimization: attractive-
ness, proximity, and exposure. Attractiveness largely refers to the value
for the potential offender of committing a crime against a specific person or
object. The proximity factor is split up into social and geographic proxim-
ity. The first relates to the degree of contact a person has with potential
offenders as a result of his or her life-style. The second has mainly to do
with residential segregation and the issue of crime and distance in that the
risk for victimization is assumed to be greater for those living in or close to
areas with many potential offenders. The greatest risk for personal victim-
ization, according to the proximity factor, is for persons who live in cer-
tain areas and move about in urban space in such way that they come into
frequent contact with potential offenders. The exposure factor is defined
by Steinmetz as "the extent to which an offender is given an opportunity to
commit an offence when he or she comes into contact with an attractive
target" (p. 65).
Hindelang, Gottfredson, and Garofalo (1978), in their theory of person-
al criminal victimization, point to the importance of life-styles: "Different
life-styles imply different probabilities that individuals will be in particular
places, at particular times, under particular circumstances, interacting with
particular kinds of people" (p. 251).
Life-style influences how often an individual is confronted with risk
situations. That is mainly assumed to be a result of how much of his or her
time a person spends in public places, especially at night. Life-style also
influences which individuals are perceived as good targets by a potential
offender. According to Hindelang et al. (1978), this has to do with judg-
ments on the behalf of the potential offender regarding the willingness of
the potential victim to report a victimization and the amount of resistance
he or she is likely to be capable of mobilizing. The similarities between the
life-style approach of Hindelang et al. and the routine activity approach of
102 5. The Urban Victim
Cohen and Felson (1979) have been pointed out many times (e.g., Garo-
falo, 1987, p. 27).
In discussing victimization, it appears important to distinguish between
cases in which the victim is a private person or not (e.g., company, etc.).
In this chapter, only crimes against private persons and their property will
be treated. This means that nonresidential burglaries, vandalism of nonpri-
vate property, and thefts of and from cars owned by companies etc. will be
excluded from the discussion. This is not to say that there are no important
factors that influence variations in victimization of company etc. property,
but just to say that the present study largely lacks relevant data to explore
this issue.
As for personal victimization, crimes may be divided into those that
actually involve the victim in the crime event (assault, robbery, and rape)
and those that normally do not (theft of and from cars and residential bur-
glaries). In the former case, life-style may be assumed to be important in
determining the degree to which individuals come into direct physical con-
tact with potential offenders (i.e., risk meetings), and in the latter case,
life-style influences to what degree the individual's property is exposed to
potential offenders (i.e., left unguarded).
Crimes of Violence
In Chapter 2, it was shown that the overall risk of victimization for violence
appears to be linked to the degree of urbanization of the area of residence.
However, it is also clear (see Chapter 8) that there are significant differ-
ences in the risk of victimization among parts of urban areas. Thus, the
victimization risk appears to be dependent on both the degree of urbaniza-
tion of an area and what parts of the urban area the individual is living in
and frequently visiting. The latter is likely to be related to life-style.
The victim-offender distinction is not always clear in crimes of violence.
Fattah (1989) states:
Who will end up being victim and who will be legally qualified as the offender
depends quite often on chance factors rather than deliberate action, planning or
intent. Thus victim/offender roles are not necessarily antagonistic but are frequent-
ly complementary and interchangeable. This is particulary true of brawls, quarrels,
disputes and altercations (p. 47).
Sex
Excluding the crime of rape, in which for natural reasons there is a strong
asymmetry between the sex of the victim and offender, the findings show
overall that except for violel1ce within the family, the victim and offender
are predominantly of the same sex (see Table 5.1). The significant excep-
tions are for minor assaults and molestations between nonfamily acquain-
tances, in which the total of cases occurring between two persons of a dif-
ferent sex comes close to half, or just above half, of all cases. For violence
occurring outside the private, the intersex cases have the highest propor-
104 5. The Urban Victim
Nonfamily acquaintances
in residences
% same sex 86 56 0 77 47
Phia
Significance n.s. n.s n.s
Public places, including
public entertainment
% same sex 91 83 0 89 59
Phi 25 20
Significance n.s. *. n.s.
Other circumstances
% same sex 78 75 0 80 65
Phi 20 35
Significance n.s. n.s. **
Note. Significance tested by chi-square test. Phi only shown in case of relationships significant
at the 5% level or better.
* Significant at 5% level.
** Significant at 1% level.
aNo female offender and no male victim in group.
Age
Generally, victims (with the exception of rape victims) are older than the
offenders and not as strongly concentrated to certain (younger) ages-
compare the age distribution in Table 5.2 with the offender age distribution
presented in Chapter 3. In crimes within the family, the victim-offender
The Victimological Approach 105
TABLE 5.2. The age distribution of victims in different legal categories of crime,a
per 10,000 in respective age group (Stockholm, 1982). b
Petty assault and Robbery
violence toward against the
Age Serious assault" public official person Raped Molestation
age correlations are strong, which is also true for nonfamily violence in
residences, with a slight exception for molestations between nonfamily ac-
quaintances (see Table 5.3). Thus, in crimes of violence in private, the
victim and the offender tend generally to be of the same age. In cases
occurring in public, the age differences between victim and offender tend
to be greater and are especially high in cases of robbery. In general, it can
also be claimed that the victim-offender age difference tends to be greater
in instrumental as compared to expressive crimes of violence. In cases of
robbery, the offender tends to be significantly younger than the victim,
whereas in rape cases (except those occurring in public) the victim tends to
be significantly younger than the offender.
Nationality
The victim and the offender tend to be of the same ethnic background (i.e.,
nationality-as dichotomized into Swedes and other nationalities). This is
of course to be expected because the population is predominantly of Swed-
ish origin. It is noteworthy, however, that the proportion of cases involving
people of same ethnicity in general is significantly lower for the rape cases.
As to the ethnic status correlations, it could generally be claimed that the
number of cases involving people with different ethnic backgrounds is high-
er in crimes of violence occurring in public than in crimes of violence in
private (see Table 5.4).
106 5. The Urban Victim
TABLE 5.3. Victim-offender age differences, mean and median age differences, and
zero-order correlation (Stockholm, 1982).
Petty assault and Robbery
Serious violence against against
assault public official Rape person Molestation
Note. Age differences calculated as victim age minus offender age. Thus, negative values of
mean and median mean that victim tends to be younger than offender (e.g., a median of-2
means that the victim on the average is 2 years younger than the offender). Correlation coef-
ficients multiplied with 100.
aSignificant at 5% level.
b Significant at 1% level.
The Victimological Approach 107
TABLE 5.4. Victim nationality, percentage of victims with foreign citizenship, and
Phi-coefficientlvictim-offender dichotomized (Swede-foreigner) nationality (Stock-
holm, 1982).
Petty assault and Robbery
Serious violence against against
Social context assault public official Rape person Molestation
Other circumstances
% same ethnicity 53 76 56 77 66
Phi 31 38
Significant n.S. ** n.S. ** n.s.
Victim age M F M F M F M F
Victim age M F M F
19 and below 25 7 16 2
20-39 51 30 56 22
40-64 35 14 41 11
65 and above 0 Ob 17 0
and in other circumstances are victimized during work (see Chap. 4).
When interpreting the figures presented in Tables 5.6 and 5.7, it should be
kept in mind that the highest victimization rates for violence are to be
found in the age group of 20 to 39 (see again Table 5.2).
property crimes considered in the present study by the fact that most thefts
of and from a private person's car occur when the car is parked nearby the
person's residence (see Chap. 4) and by the obvious link to the victim's
area of residence for the crime of residential burglary.
Just as for crimes of violence, the present study shows that the risk of
victimization for property crimes is strongly related to degree of urbaniza-
tion of area of residence (see Chap. 2). However, as for crimes of violence,
the risk of victimization for property crimes appears to fluctuate strongly
depending on the victim's area of residence in larger urban areas (see
Chap. 8). There is no significant association between victimization rates for
violent crimes and property crimes at the city-area level. Thus, the victi-
mization rate for property crimes shows a pattern different from that for
violent crimes. Moreover, in contrast to crimes of violence, there are some
marked differences in victimization risk by area of residence among diffe-
rent types of property crimes (see further Chap. 8).
Conclusion
At the risk of oversimplification, one may say that the prime overall risk
for violent victimizations is strongly dependent upon what kind of people
one associates with and what places (activities) one frequently visits. The
prime overall risk for victimization of the types of property crime included
in this study appears to be largely dependent upon area of residence.
However, this is not to say that individual behavior is unimportant for
property crime victimization risks, only that area of residence appears to
be a major risk factor and that we lack data to put light on the importance
of individual behavior.
6
Stockholm: Its Crime and
Urban Structure
111
112 6. Stockholm: Its Crime and Urban Structure
TABLE 6.1. Population in the Greater Stockholm area and, separately, for the city
of Stockholm and Greater Stockholm area ouside the city, 1950-1985.
Area 1950 1960 1970 1980 1985
caD
Inner-city
[ill Post-war sleep. words
~ Post-wor ABC words
III Post -wor Jarvo cHy
Single femily houses ..
D other ou ter city
* 5 I :i or more.
Not i nc1.
MAp 6.1. Key map of the research area showing the location of the CBD , the inner
city, major post war residential developments, and single-family housing wards.
Crime Rates and Crime Trends in the City of Stockholm 113
the present study covers the inner-city area and those suburbs belonging to
the city of Stockholm (see Map 6.1). In general, one can say that these are
the suburbs surrounding the inner city; the more distant suburbs are not
included in the study. Hereafter, Stockholm will be used to mean the city
of Stockholm, and the conurbation will be referred to as the Greater Stock-
holm area.
The somewhat arbitrary administrative boundaries between the research
area and the rest of the Greater Stockholm area might cause some prob-
lems in an ecological study because the included suburban areas do not
have any distinct natural border with the neighboring, not-included sub-
urban areas. It is, however, not believed that this fact will distort the gener-
al analysis of the relationship between various types of urban environments
and different aspects of victimization and crime, even though for special
kinds of analysis, the problem of the somewhat arbitrary borders might be
of greater importance-for instance, in the analysis of crime and distance
(Chap. 8)..
TABLE 6.2. Total crimes against the penal code and selected types of crimes; num-
ber and per 10,000 inhabitants; 1951, 1960, 1970, 1980, and 1985 (City of Stock-
holm).
Crime 1951" 1960 1970 1980 1985
TABLE 6.3. Regression coefficients and R2, 1968-1987; crime rates and increase
1968 to 1987 for crimes against the person and vandalism; mean crime rates 1968-
1973, 1974-1980, 1981-1987 for serious property crimes; all figures per 10,000 resi-
dents.
Greater Stockholm The counties Other Sweden
City of area (excluding of Goteborg (non-big-city
Stockholm city of Stockholm) and Malmo areas)
incl.
MAP 6.2. Wards with greater industrial areas and commercial centers.
0 .31 -
. 0.21 - 0 .30
0 . ' , - 0 .20
[ ] 0 .0' - O. , 0
D 0 .00
inc\.
i ncl.
major shopping area, and at night it is the central part of the major public
entertainment district, which, however, spreads into nearby parts of the
inner city (see Map 6.3). During both daytime and nighttime , the great
majority of people in the area are nonresidents.
the 1950s and 1960s; and (3) the Jarva city, built in the late 1960s and early
1970s (see again Map 6.1). These developments were closely linked to the
development of the Stockholm underground, which was begun in 1945 and
largely completed in the mid-1970s.
The major difference between the sleeping suburbs and the ABC sub-
urbs is that the latter were planned not only to be residential areas, built
around tube stations, but also to include a substantial number of work-
places and a large cultural and commercial center. ABC stands for work,
residence, and center (compare Maps 6.1 and 6.2). The Jarva city is the
most recent major residential development within the boundaries of the
city of Stockholm. It includes the most densely built and densely populated
residential areas outside the inner city (Johansson, 1987, p. 596). One
salient characteristic of the population of Jarva city is that it includes two
wards with, for Swedish conditions, comparatively very high proportions of
immigrants (40-50%).
Housing in Stockholm
A rather small proportion of the inhabitants of Stockholm live in single-
family houses: around 10% in the whole city and around 20% in the sub-
urban areas. Most residents live in rented flats, mostly rented from private
landlords, but a substantial proportion of the rented flats are also owned by
nonprofit organizations (see Table 6.4).
Because the overwhelming majority of the one- and two-family houses
TABLE 6.4. Housing in the city of Stockholm; number and percentage of resi-
dences by type of housing (1980).
Type of housing Number %
Owner-occupied
House a 38,797 10.6
Tenant-owned flat (condominium) 63,438 17.3
Rented
Private landlordb 160,437 43.8
Nonprofit organization c 98,972 27.0
State, municipalityd 4,748 1.3
Total 366,392 100.0
Missing 797
aOne- or two-family houses owned by private person (37,652) or tenant owners' association
(1,145).
bMultistory houses owned by private person or company (including 327 residences in one- or
two-family houses).
crncIuding 877 residences in one- or two-family houses.
dIncIuding 252 residences in one- or two-family houses.
Source: Statistisk firsbok fOr Stockholm 1983/84, USK Table 68 (p. 121).
The Factorial Social Ecology of Stockholm 121
DomineUng type of
tenure (51 ~ or more)
Not Incl.
MAP 6.4. Dominating type of tenure (51 % or more) in the wards of Stockholm,
1980.
Familism
The first, and strongest, factor was interpreted as Familism, with high posi-
tive loadings on the variables of household size, households with children!
youth, persons aged 0-14 years, fiats in one- and two-family houses, large
residences, distance from city center, and mean income-and with high
negative loadings on the variables of small residences, persons aged 60 or
more years, overcrowded fiats, children with single working parents, the
percentage of females of those aged 15 years and older, and old residences.
An alternative labeling of this factor may be Familism/Space, because
obviously some of the high loading variables are of relevance for space.
The plotting of the factor scores in five classes is shown in Map 6.5. Most
The Factorial Social Ecology of Stockholm 123
TABLE 6.5. Factor loadings and communalities. Varimax rotated factors, Stock-
holm, 1980 (N = 125).
Variable II III IV V h2
Contribution % 37 21 10 6 6
Note. Coefficients have been multiplied by 100; loadings 40 and higher in bold type.
inner-city wards show negative scores for this factor, whereas the highest
positive scores for this factor are to be found among parts of the postwar
multistory residential developments and single-family house areas. One
may hypothesize that mainly less better-off families with children are
housed in the newer multistory residences (often nonprofit public hous-
ing), and that the better-off families with children are mainly housed in
single-family house areas. In general, Familism is highest in the outskirts of
the outer city.
124 6. Stockholm: Its Crime and Urban Structure
FAMILISM .
Fllctor scores;
1.91 - 2.88
09
.3 - 1.90
-0 .05 - 0.92
[ ] - 1.03 - -0 .06
D -2 .0 I - -1.04
i ncl.
; nc1.
PROBLEM AREAS
Factor scores ;
2.97 - 4. 14
1.79 - 2.96
0.61 - 1.78
[] -0 .57 - 0.60
D -1.75- -0.58
i ncl.
Not incl .
problem areas are not of the same kind as the ones producing high scores
on this factor in the outer city. In this connection, it should also be pointed
out that the high-scoring CBD ward belongs to the group of wards with the
smallest residential population, which may make the figures for this area a
bit shaky.
When speaking about problem residential areas, it should be pointed out
that what may be considered a problem residential area is relative and that
many of the high-scoring areas in Stockholm probably not would qualify
for the label "problem area" in an international perspective. There are no
slum areas in Stockholm as there are in many big U.S. cities. Moreover, it
is also questionable whether Stockholm's problem residential areas are
comparable (on the level of social problems) to inner-city run-down areas
and so-called difficult council estates in large British cities (see Baldwin &
Bottoms, 1976).
Social Rank
The third factor was interpreted as Social Rank, with high positive loadings
on the variables of income heterogeneity, mean income, workplaces in
commerce, and vacant flats-and high negative loadings on the variables
of blue-collar workers and distance from city center.
126 6. Stockholm: Its Crime and Urban Structure
SOCIAL RANK .
Fector scores ;
3.06 - 4.37
1.74 - 3.05
0.42- 1.73
ITJ -0.90 - 0.4 1
D -2 .22 - -0.91
inel.
inel.
Looking at the plotting of the scores for the Social Rank factor (see Map
6.7), it is obvious that there are two areas of high social rank in Stockholm:
the northern part of the inner city, especially the CBD and the areas close
to the CBD, and the single-family house area close to the inner city in the
west of the outer city. This result corresponds very well with a Stockholm
resident's everyday view of the most prestigious areas to reside in.
Workplace Areas
The fourth and final interpreted factor was labeled Workplaces because it
had high positive loadings on the variables of percentage of workplaces,
manufacturing workplaces, and workplaces in commerce. It is a factor that
picks out those wards having a high frequency of workplaces (see Map 6.8,
and compare Map 6.2).
WORK PLACES
Fector scores ;
2.85 - 4.00
1.69 - 2.84
0 .53 - 1.68
lTI -0.63 - 0 .52
0 -1.79 - - 0 .64
i ncl.
Not i ncl.
mean income (64) and blue-collar workers (-84) being higher (compare
Table 6.3).
None of the interfactor correlations was above .31. Familism was nega-
tively related to Workplace (- .31), Problem Residential Areas negatively
related to Social Rank ( - .28), whereas Social Rank was negatively related
to the noninterpreted Factor V ( - .26) . The decision was taken to stick to
the orthogonal solution in light of the generally low or modest intercorrela-
tions and the more complex interpretation that results when using corre-
lated factors . The purpose of the analysis presented in this chapter is purely
descriptive , and for that purpose the Varimax solution was judged to give
an adequate picture of the Stockholm urban structure.
CBD (I)
Inner high ronl< (2)
.Otherlnner(3)
Outer hi gh rllnl< (4)
III Outer problem (5)
D Other outer (6)
Not i nc1.
When grouping the 123 wards included in the crime study into major
types of urban environments, three principles of grouping were chosen:
1. Separate inner- and outer-city groups of wards.
2. Try as far as possible to have geographically coherent groups of wards.
3. Use the factor scores for the Problem Residental Area and Social Rank
factors as the main criteria for grouping within the inner-city area re-
spective to the outer-city area.
This grouping resulted in six major types of urban environments, which
consist of 15 subgroups when the criteria of geographical coherence is up-
held. The six major types are as follows:
Inner City
1. The central business district (CBD).
2. Inner-city high social rank areas.
3. Other inner-city areas.
Outer City
4. Outer-city high social rank areas.
5. Outer-city high problem residential areas.
6. Other outer-city areas.
The locations of the different types of major urban environments are
shown in Map 6.9. Selected characteristics of the major urban environ-
ments are shown in Table 6.6.
7
Housing, Population Composition,
and Offending
130
Housing, Population Composition, and Offending 131
1. The claim that the relative area distribution of delinquency rates in Chicago is
stable over time has recently been challenged by Bursik (1986b) in a reanalysis of
the Shaw & McKay data for the period 1930-1970.
The Offender Residential Distribution 133
2. The offender sample in Baldwin & Bottoms's (1976, p. 66) study was a sample of
all offenders convicted or cautioned in 4 different months in 1966. It is unclear to
this author whether the rate figures presented for Sheffield areas are weighted to
I-year figures or only represent the rate for the 4 months included.
134 7. Housing, Population Composition, and Offending
Second, the fact should be brought out that the presented rates are not
pure participation rates because the same offender might appear several
times. As discussed in Chapter 3, the aggregate offender rate, for instance,
for a ward, can be viewed as produced by the residents' participation in and
their individual frequency of crime. For cases in which the offender series
for violence and vandalism is shown separately, the offender rates are very
close to the participation rates, because most offenders are only included
for one crime of these kinds. The offender rates for serious property crimes
is to a somewhat higher degree influenced by individual frequency, because
there are a number of offenders included who appear in the study with
several crimes of this kind. However, by and large, given the short period
of observation, the rates presented are close to pure participation rates,
because the bulk of offenders are included only for one crime. As for rates
for single wards, especially for the smaller of the wards, the offender rate
might be significantly affected by, for instance, cases of serial clearance.
This fact will be commented upon when applicable in the analysis carried
out below. Some results presented below indicate that areas with high par-
ticipation rates also have high individual frequencies of crime among offen-
ders and that areas with low participation rates have lower individual
frequencies of crime.
Third, the basic offender rates will be given per 1,000 residents aged 10
or older. There are, of course, several possibilities in choosing a denomina-
tor for the calculation of offender rates. For instance, one can always argue
that because rather few offenders are in older age groups (in this study 5%
of the total offenders are 50 years or older, and 2% are 60 years or older),
one should not only restrict among the youngest ages to be included in the
denominator, but also among the oldest ones. However, because the pro-
portion of victims in crimes involving older persons is much higher than
the corresponding proportion of offenders (see Chap. 5), and because I pre-
ferred to have comparable rates of offending and victimization, I decided
not to leave out the older ages from the denominator when calculating
the offender rate. To some extent this problem is dealt with anyway be-
cause age-standardized rates will also be calculated and included in the
analysis.
Fourth, it has rightly been pointed out that when studying the residential
distribution of offenders in a city, with crimes committed in the city as
sampling frame, one leaves out those offenders who live outside the city
but commit crimes in the city and those offenders who live in the city but
commit crimes outside the city (Baldwin & Bottoms, 1976, p. 72). Only the
latter can be considered to be a real problem in the study of the offender
rate variation in the urban area, and only insofar as these offenders are
distributed differently from those who commit crimss within the urban
area. This problem will be further discussed in Chapter 8.
Finally, the problem of possible bias in studies using officially recorded
crimes, due to area differential dark figures and clearance, was brought out
early in the discussion of the Shaw and McKay (1969) findings. Mostly, it
The Offender Rate Distributions 135
seems to have been assumed that the offender rate differences noted in
studies of recorded crimes overestimate the area differences. (For a gener-
al discussion on class bias in official data, including some area aspects, see
Braithwaith, 1979 pp. 32-46) However, there are few empirical studies
showing this to be the case. On the contrary, the few conducted generally
do not support this view. For instance, Bottoms and Wiles (1986), summar-
izing the efforts in the Sheffield project to explore this possible bias, state
that:
The results suggest that a good deal of confidence can be placed on the assumption
that within a city the residents of small areas that have high official (police re-
corded) offender rates usually do .actually commit more offenses than do the resi-
dents of areas with low police-recorded offender rates. (p. 107)
The same view was also held by Shaw & McKay (1969, p. 45) based on
their extensive study of Chicago male juvenile offenders.
TABLE 7.2. Offender rates per 1,000 residents aged 10 or more (except youth offen-
ders, which shows rate per 1,000 residents aged 10-19) (Stockholm, 1982).
Range Median %
excluding excluding wards with Extreme
Series extremes extremes zero rate Skewness values
Extreme Values
The extreme values were picked out for special inspection to see whether
these were produced by, for instance, serial clearance or other similar fac-
tors. As extreme values were defined those that were clearly out of line
with the others when plotting the ward rates. For instance, considering the
plot of rates for total offenders ordered by size (see Fig. 7.1), one can
clearly see that there is a rather continuous increase up to around 35 offen-
ders per 1,000 residents aged 10 or older, when suddenly the rate makes a
significant jump for the last two wards with the highest frequencies. It was
found that all extreme values but one-that for the youth offenders (see
again Table 7.2)-belonged to the same two wards: one inner-city CBD
ward and one outer-city ward. In the latter case, it was found that the
offender rate for this ward was highly affected by a serial clearance case of
a serious property offenders, and therefore the offender rate for this par-
ticular ward might be misleadingly high. For no other ward, including the
CBD ward with extreme values for some of the offender rates, was it possi-
ble to detect any case of single offenders that significantly affected the
offender rate. The extreme offender rates noted for the CBD ward might,
due to its small population size, be affected by chance variations. The ex-
treme value for youth offenders might also be affected by chance variations
because it was localized to a ward with a small youth population (88 resi-
The Offender Rate Distributions 137
70
60
so
40
30
20
10
0
0 so. 100 ,
FIGURE7.1. Plot of ward offender rates for total offenders, in order by size of rate
(N= 123).
dents aged 10 to 19). It was decided to exclude the extreme values from the
further analyses of the offender distributions.
TABLE 7.3. Segregation indexes and Gini coefficients for different offender dis-
tributions (Stockholm, 1982).
Segregation Gini
Series indexes) coefficient" N
'Figures within brackets show Gini coefficients when extreme values were included. As can
be seen, the effect is marginal.
3. The Gini coefficient is a measure similar to the segregation index (see Duncan &
Duncan, 1955a, p. 211). The minimum value of the Gini coefficient is the value of
the segregation index(s), and its maximum value is 2s - S2. The correlation be-
tween the two measures is high.
The Offender Rate Distributions 139
100
90
....0)
III
"tJ
c: 70
....
....0
0)
...c:
0)
"....
0)
50
c.
0)
>
.;;
40
.!!!
;j
E
;j
u 20
10 20 30 40 50 60 70 80 90 100
Cumulative per cent population
100
90
~ 80
0)
0
c:
......
0) 70
0
~
c:
60
0)
" 50
~
Q)
>
;::; 40
~
:I
E 30
:I
:.J
20
10
10 20 30 40 50 60 70 80 90 100
Cumulative per cent population
FIGURE 7.3. Lorenz curves for chronic and youth offenders, respectively (Stock-
holm, 1982).
140 7. Housing, Population Composition, and Offending
100
Vandalism
90
80
...'"
Q)
-c 70
c:
....
....0
Q)
...c: 60
Q)
... 50
(.)
Q)
c.
Q) 40
>
.;::;
.!!!
:J
E
:J
U
10
10 20 30 40 50 60 70 80 90 100
Cumulative per cent population
FIGURE 7.4. Lorenz curves for violent, vandalism, and serious property offenders,
respectively (Stockholm, 1982).
with that problem is to look at the so called Lorenz curve, which adds the
information of what kind of distribution lies behind the coefficient.
The Lorenz curves for the six different offender distributions are shown
graphically in Figures 7.2, 7.3, and 7.4. As in Chapter 3, when describing
the offender concentration in a cohort, the normal way of presenting the
Lorenz curve has been mirrored, following the praxis employed by Fox and
Tracy (1988).
In comparison to having individuals as units of analysis (as in Chap. 3),
having unequally sized wards as units makes the interpretation of the
Lorenz curve (and Gini coefficient) less straightforward. Moreover, other
ward divisions may yield more or less different distributions and coef-
ficients, a fact that makes intercity comparisons of the degree of offender
concentration somewhat complicated. It appears that area offender con-
centration comparisons (using segregation indexes, Gini coefficients, or
similar measures) are most straightforward when comparing different
offender distributions in the same city or changes in a single city over time,
assuming, in the latter case, that the ward division is the same and that
changes in the population distribution have not been too large.
All in all, the result of the study of offender area concentration shows
quite a high degree of concentration. This is further illustrated in Table
7.4, in which the percentage of offenders by quartiles of the population
ordered by rank of ward offender rate is shown. It might be worth noting,
The Offender Rate Distributions 141
aFor example, for total offenders, the first category, upper quartile, may be read as that the 22
wards with the highest offender rates in the city have 25% of the population but 46% of the
offenders.
with due caution as regards all the difficulties of intercity comparisons, that
the offender area distribution in Stockholm 1982, by quartiles of popula-
tion, is remarkably similar to that found by Shaw and McKay (1969) for
male juvenile offenders in Chicago in the early part of this century. Shaw
and McKay (p. 71) found, for instance, for their 1927-1933 series, that
54.3% of the delinquents belonged to the upper quartile, 23.9% to the
second quartile, 14.6% to the third quartile, and 7.2% to the lower quartile
of the population when areas were ranked by rate of delinquents. How-
ever, the fact that the pattern of the area concentration appears to be simi-
lar, should not be confused with similarity in rates of offending, which
is a totally different matter.
TABLE 7.5. Range, mean, and median ward offender rates for subcategories of
offenders by quartiles of ward total offender rates; zero-order correlations between
total offender rate and offender rates for series of subcategories of offenders
(Stockholm, 1982).
Serious
Total offender Chronic Youth Violent Vandalism property
ward rate offenders offenders offenders offenders offenders
Upper quartile
Range 1.5-22.3 0.0-111.8 0.0-13.4 0.0-13.3 0.0-25.6
Mean 9.8 31.1 8.3 2.8 15.3
Median 9.5 18.6 8.5 2.2 15.3
N 30 30 30 30 30
Second quartile
Range 0.0-19.9 0.0-131.7 0.0-12.2 0.0-19.9 0.0-17.1
Mean 5.2 22.2 5.8 1.9 8.3
Median 4.2 11.2 3.3 1.3 5.8
N 31 30 31 31 31
Third quartile
Range 0.0-7.0 0.0-123.5 0.0-10.4 0.0-6.7 0.0-9.0
Mean 1.8 14.8 4.1 1.7 2.3
Median 1.3 6.4 3.5 1.2 1.7
N 30 30 30 30 30
Lower quartile
Range 0.0-4.1 0.0-27.2 0.0-5.0 0.0-2.2 0.0-3.8
Mean 0.5 3.3 1.4 0.1 0.3
Median 0.0 0.0 0.0 0.0 0.0
N 30 30 30 30 30
r (N = 121)a 0.72 0.36 b 0.70 0.36 0.87
ders, although significantly correlated with the total offender rate, shows
a much lesser degree of association (see again Table 7.5). This finding
should be interpreted with some care, because the two subseries of offen-
der categories showing the least association with the total offender rate are
those series that may be most affected by chance variation, as discussed
earlier in this chapter. Recalculating the correlation of the youth offender
series with the others, eliminating the 41 areas with fewer than 200 resi-
dents aged 10 to 19 resulted, as expected, in an increase in the magnitude
of the correlation (i.e., r = 0.46, p = 0.000, n = 81).
A comparison of the overlap of the wards in the upper quartiles of,
respectively, the five subseries of offender categories to the wards in the
upper quartile of the total offender rate shows the following percentages
The Offender Rate Distributions 143
of the wards in the highest quartile of the respective series to also belong to
the wards of the highest quartile for the total offender rate:
Chronic offenders, 70%.
Youth offenders, 43%.
Violent offenders, 63%.
Vandalism offenders, 47%.
Serious property offenders, 80%.
The intercorrelations among the subseries of offender categories are
shown in Table 7.6. It is worth noting that the subseries for youth and
chronic offenders shows no significant relationship at all. That also holds
true when recalculating the relationship to exclude the 41 areas with a
small youth population (i.e., r = 0.21, N.S., n = 81). Moreover, comparing
the overlap of the number of wards in the upper quartile of chronic and
youth offender rates shows the number of overlapping wards to be 9 out of
30 (i.e., 30%). In this connection, it is also worth mentioning that the
individual overlap between youth and chronic offenders is minimal-less
than one half of 1% of the chronic offenders are under the age of 20.
33 -
25 - 32
17 - 24
[] 9 - 16
o 0 - 8
inel.
Not inel.
MAp 7.1. Total offender rate per 1,000 residents aged 10 or older (Stockholm,
1982).
incl .
MAP 7.2. Chronic offenders per 1,000 residents aged 10 or older (Stockholm,
1982).
The Offender Rate Distributions 145
Not i ncl.
MAP 7.3. Youth offenders per residents aged 10-19 (Stockholm, 1982).
9-
.7-8
5-5
8'.01
t:.;ij 3 - 4
D 0-2
Not i ne1.
MAP 7.4. Violent offenders per 1,000 resident's aged 10 or older (Stockholm,
1982).
146 7. Housing, Population Composition, and Offending
10 yeers or 01 der.
8 -
6 - 7
4-5
EJ 2 - 3
0 0- 1
; ncl .
MAp 7.5. Vandalism offenders per 1,000 resident's aged 10 or older (Stockholm,
1982).
of the cities (e.g., Mays, 1963, p. 160; Baldwin, 1975; Baldwin & Bottoms,
1976). Herbert (1978) comments, comparing the high offender rate area
patterns in the United States and Great Britain, "The basic distorting fac-
tor would appear to be the public sector intervention in the British housing
market , which in the process of transferring population groups from city
centre to periphery has also transferred their behaviour patterns" (p. 235).
All in all, there is no universal pattern of the location of high offender
rate areas, and this pattern may also change over longer periods of time.
(See also Herbert , 1972, p. 217, and Harries, 1974, pp. 74-75, for exam-
ples from countries other than those discussed here.)
Turning to Stockholm, 1982 (excluding the extremes from considera-
tion), most of the wards in the two highest classes of the total offender rate
are to be found in the outer-city areas, but some of the wards in the south-
ern part of the inner city also belong to the high offender rate wards (see
Map 7.1). Generally, the same overall pattern holds for the different
subseries of offenders (see Maps 7.2-7.6), although, as has already been
noted, when exploring the relationship between these distributions, the
high offender rate wards of respective series show some differences in
distribution.
Age Structure and Offender Rates 147
incl.
MAp 7.6. Serious property offenders per 1,000 residents aged 10 or older (Stock-
holm, 1982).
4. Only offenders residing in the city were included in calculating the p-values.
Offenders aged 60 and older were referred to as one class.
148 7. Housing, Population Composition, and Offending
n3, etc.). Third, the expected number of offenders for a given ward (EO)
was computed as follows:
EO = n1 x PI + n2 X P2 + n3 X P3 + ... nk X Pk
Fourth, to get the expected offender rate (ER) for a ward, the expected
number of offenders (EO) was simply divided by the residential population
aged 10 or older. Finally, the difference between the expected offender
rate (ER) and the observed offender rate (OR)5 was taken, and the result-
ing residual was used as a measure of the age-standardized offender rate
(ASOR):
ASOR=OR-ER
The AS OR will be positive if the ward has a higher observed than expected
offender rate and negative if the ward has a lower observed than expected
offender rate. The ASOR will subsequently be used in some of the analyses
below as a measure of the offender rate when controlling for area age struc-
ture.
The zero-order correlations of the expected offender rates (ER) with the
observed rates (OR) are shown for the different offender series in Table
7.7. The low and, with one exception, nonsignificant correlations between
the expected and observed offender rates show that the offender rate varia-
tions in Stockholm cannot very well be explained by area differences in age
structure. In line with this, the zero-order correlations between the age-
standardized (ASOR) and observed offender rate (OR) series are very
high (i.e., for total offenders, .94, for chronics, .97, for violent offenders,
.97, for vandalism offenders, .99, and for serious property offenders, .96).
As an illustration, the scattergram and regression line for the relationship
between OR and ASOR for total offenders is shown in Figure 7.5. All in
all, it seems safe to conclude that variations in the age structure of the
wards are no major cause of the area offender rate variations.
TABLE 7.7. Zero-order correlations of observed and expected offender rates per
1,000 residents aged 10 or older (Stockholm, 1982).
Observed offender rates (OR)
Serious
Expected offender Chronic Violent Vandalism property
rates (ER) Total offender offender offender offenders
Age-standardized
total offender rate
20
.,
-2
-25
-30
0 5 j 0 15 20 25 30 35 40
Total offender rate.
FIGURE7.5. Scattergram and regression line for relationship between OR and
ASOR for total offenders.
England (e.g., Baldwin & Bottoms, 1976; Bottoms & Wiles, 1986) have
shown that there are great area-level differences in offender rates by type
of tenure dominating an area. These studies used a classification of type of
tenure of area in four classes: (1) owner-occupied, (2) rented from private
landlord, (3) rented from local authority (public housing), and (4) mixed.
If more than 50% of the households in the area belonged to one of the first
three classes, the area was referred to that class, otherwise as mixed. Bot-
toms and Wiles found no owner-occupied area with a high offender rate,
but in the two other main types of areas they found that "some privately
rented districts had high rates, and some had low rates; some local author-
ity areas had high rates, and some had low rates" (p. 105). Their main
explanation for these differences is the mechanisms of housing allocation;
they state that "the mechanisms of allocation between and within tenure
types must be a critical part of any explanation of offender-rate distribu-
tion" (p. 119). For instance, Bottoms and Xanthos (1981) suggest, from
the Engiish experience, the following possibly important factors influenc-
ing offender rate variations in the public sector of housing: (1) differential
rent levels, (2) local authority allocation systems, (3) residents' self-
selection to area, and (4) area reputation.
It has also been maintained that "the way we design our apartment
buildings and housing complexes has everything to do with the crime pat-
tern of the city" (Jeffery, 1977, p. 193). Physical design characteristics of
buildings and housing estates may have an effect on residential area offend-
ing rates by their influence on social space, privacy, and surveillance
opportunities-all factors that have been assumed to be of relevance for
social life and social control in residential areas (e.g., Newman, 1972;
Baldwin & Bottoms, 1976; Brown & Altman, 1981).
In this section, I shall explore the relationship between housing and
offending. Housing in Stockholm and the distribution of types of housing in
areas were presented in Chapter 6. The housing classification used in the
following analyses was also presented in Chapter 6. Before turning to the
1982 Stockholm data, I shall summarize some findings from a study made
on the individual-level relationship between housing and offending in
childhood and youth in a Greater Stockholm area cohort.
TABLE 7.8. Participation rates by type of housing in childhood and youth for
traditional crimes; the individual-level relationship (Project Metropolitan, 1953
Greater Stockholm cohort).
Total cohort a Males b Femalesc
Owner-occupied
House 1.4 7.5 2.5, 13.0 0.2 1.7
Flat 0.5 7.6 1.1 13.8 0.0 1.5
Rented
Privately 2.2 9.8 3.3 15.9 1.0 3.3
Nonprofit 3.2 14.0 5.5 23.5 0.8 4.1
Other public housing 2.4 11.6 3.9 19.2 0.7 3.7
Significance
(prob. value) 0.000 0.000 0.000 0.000 0.012 0.000
aN= 13,855.
bN=7,083.
cN=6,772.
Note. Highest participation rate in bold type.
remarks are necessary. First, this study is restricted to the relationship be-
tween housing and offending in childhood and youth. As previously dis-
cussed, the relationship between housing and offending may be different
for different age groups. Second, the area studied covers the Greater
Stockholm area and is not confined to the city of Stockholm (see Chap. 6).
Participatio n
Both in youth and in childhood, both totally and separately for the two
sexes, the results show significant differences in participation. 7 Generally,
the highest participation rates were found among those living in the rented
sector of housing, and particularly among those living in flats rented from
nonprofit organizations, that is, public housing (see Table 7.8).
Frequency of Crime
The comparison of frequency of crime by type of housing was only possible
for the youth period (13-19 years) because frequency data were lacking for
6. For the full details and other findings than those presented here, the reader is
referred to the original study. The Project Metropolitan cohort study is presented
in Chapter 3 of the present study.
7. Only results for traditional crimes (i.e., violence, vandalism and theft) will be
shown here, although the full range of crimes was included in the study (see
Wikstrom, 1989).
152 7. Housing, Population Composition, and Offending
Owner-occupied
House 5.2 5.6 1.8
Flat 4.3 4.6 2.0
Rented
Privately 5.7 6.4 1.7
Nonprofit organization 8.3 9.2 2.5
Other public housing 4.9 5.3 2.1
Difference (times)
highest -lowest 1.9 2.0 1.5
the childhood period. The result shows that as regards frequency of crime
there are also marked variations according to types of housing, although
less so for females (see Table 7.9). Again, those living in nonprofit housing
have the highest values. All in all, participation in and frequency of crime
tend to covary when compared by type of housing.
TABLE 7.10. Range, mean, and median offender rate by type of housing dominat-
ing area (Stockholm, 1982).
Serious
Dominating Chronic Youth Violent property
housing- Total offenders offenders offenders Vandalism offenders
Owner-occupied
House
Range 0.0-21.0 0.0-7.0 0.0-131.8 0.0-10.4 0.0-5.7 0.0-18.1
Mean 5.6 1.0 19.5 1.8 0.9 2.8
Median 5.5 0.0 0.0 0.0 0.0 0.0
N 29 29 29 29 29 29
Rented
Privately
Range 0.0-28.2 0.0-22.3 0.0-98.0 0.0-12.3 0.0-19.9 0.0-22.3
Mean 12.3 4.1 9.9 4.5 1.9 5.6
Median 10.4 2.4 0.0 3.9 0.0 2.9
N 45 46 47 47 46 46
Non-profit org.
Range 4.1-35.7 0.0-18.9 0.0-123.5 0.0-13.4 0.0-9.5 0.0-25.6
Mean 22.4 7.8 28.2 8.8 2.3 11.1
Median 24.5 7.4 20.6 9.1 1.8 11.7
N 24 24 24 24 24 24
Mixed
Range 2.1-32.6 0.0-20.5 0.0-111.8 2.1-11.4 0.0-5.7 0.0-23.0
Mean 16.0 6.1 21.9 5.7 1.3 9.0
Median 15.9 4.6 8.5 4.7 1.3 8.1
N 20 20 19 20 20 20
the findings show the same pattern as that in the individual-level analysis,
namely, that the areas dominated by rented housing have higher offender
rates than those dominated by owner-occupied housing, and that among
the areas dominated by rented housing, those where the majority of hous-
ing is nonprofit have the highest median offender rates (see Table 7.10).
This holds true for total offenders and all subseries of offender categories,
with the minor exceptions that the median offender rates for youth and
vandalism offenders are 0 both in areas dominated by owner-occupied
houses and those dominated by housing rented from private landlords.
Moreover, the findings show, just as for Sheffield, that the offender rates
vary significantly among areas dominated by the same type of housing. As
regards the finding in the Sheffield study-that no owner-occupied area
had a high offender rate-if we define a high offender rate area as one
belonging to the upper quartile of the area offender rates, then there are
154 7. Housing, Population Composition, and Offending
Zero-Order Correlations
A zero-order correlational analysis, with dominating housing in an area
represented by a set of dichotomous variables, generally shows a negative
association with the offender rate for owner-occupied housing areas and a
positive association for nonprofit housing areas. The subseries for vandal-
ism and youth offenders are exceptions in that vandalism shows no signif-
icant correlation to either owner-occupied housing or nonprofit housing,
and youth shows no significant correlation to owner-occupied housing.
Mixed areas and areas dominated by fiats rented from private landlords
Serious
Total Chronic Youth Violent Vandalism property
Dominating housing offenders offenders offenders offenders offenders offenders
Multiple Regression
The analysis presented above shows, generally, that owner-occupied
(negative) and nonprofit housing (positive) are those types most strongly
linked to area offender rate. To explore their joint capability to explain
variations in area offender rates, multiple regression analysis was carried
out, with the percentage of households living in flats rented from nonprofit
organizations and the percentage of households living in owner-occupied
houses as independent variables. 8 The result shows that type of housing
dominating area to a rather high degree (except for the subseries of youth
and vandalism offenders) explains the area variation in offender rates (see
Table 7.1~). However, this finding might just reflect the residential seg-
regation by type of housing for different social groups. I shall return to that
question after having explored, in the next section, the relationship be-
tween population composition and offender rates in the areas of Stock-
holm.
TABLE 7.12. Multiple R2, standardized partial regression coefficients, and probabil-
ity values; housing and offender rates (Stockholm, 1982).
Dependent variable
Serious
Independent Chronic Youth Violent Vandalism property
variables Total offenders offenders offenders offenders offenders
% owner-occupied houses
Standard regular
coefficient -.27 -.21 .07 -.34 -.17 -.12
Probability value 0.00 0.02 N.S. 0.00 N.S. N.S.
% nonprofit housing
Standard regular
coefficient .52 .40 .30 .43 .03 .48
Probability value 0.00 0.00 0.00 0.00 N.S. 0.00
Note. Number of observations varies between 121 and 123 depending on offender series in-
volved. N.S. = Nonsignificant at the 5% level.
8. Adding the percentage of privately rented fiats to the equation did not at all
improve the variance explained.
156 7. Housing, Population Composition, and Offending
9a. The reader is reminded that the oblique solution (Promax) showed the factors to
be only weakly correlated (see Chap. 6).
Population Composition and Offending 157
ism variable was 0.96, and for the major Social Problem household vari-
able .92. 9 The three chosen independent variables are defined as follows:
1. Mean household size (familism).
2. Percentage of households receiving public welfare assistance (social
problems).
3. Percentage of blue-collar workers (SES).
The use of households receiving public welfare as a measure of the pre-
sence of residents with social problems may require some justification. It is
acknowledged that this is a crude measure. The two major categories re-
ceiving public welfare assistance are families with children having monet-
ary problems and socially marginalized persons (e.g., drug addicts and
alcoholics). The assumption made is that the proportion of families receiv-
ing public welfare in an area will reflect the proportion of households with
different kinds of social problems in that area.
The zero-order correlations among the three independent variables
shows that mean household size of area is virtually unrelated to the percent-
age of households that receive public welfare assistance (r = - .05) and to
the percentage of blue-collar workers (r = .01), while the percentage of
blue collar workers and the percentage of those receiving public welfare
assistance, as may be expected, shows a fairly high correlation (r = .54),
although not of the magnitude to cause any serious problems of multi-
colinearity. However, an inspection of the scattergrams of the interrela-
tionship of these variables (Figures 7.6, 7.7, and 7.8) show that it may be
too hasty to say that mean household size is unrelated to the percentage of
those receiving public welfare assistance (see Fig. 7.6).10 There is a group
of wards among those with the lowest level of residents with public welfare
assistance that have the highest scores for familism. The group consists
mainly of outer-city, high-SES, owner-occupied housing areas (see Chap.
6). If wards with less than 3% of the households receiving social welfare
are excluded, there is a strong correlation (r = .80, p = .000, N = 56) be-
tween mean household size and the percentage of those receiving public
welfare assistance. l l So it appears, that this relationship changes with the
area level of residents receiving public welfare assistance-a fact to be
taken into consideration in the analyses to come.
9. All regression analyses, presented below, were also made with factor scores as
independent variables, but the results were nearly identical, although the factor
scores generally explained a few per cent more variation in the multiple regression
analysis.
10. A ploynominal regression analysis of mean household size and per cent receiv-
ing public welfare assistance showed a clearly significant relationship (p = 0.001,
R2 * 100 = 22.9).
11. The relationship between the two variables when only wards with less than 3%
receiving public welfare assistance (N = 67) were included is negative, and nonsig-
nificant at the 5% level.
158 7. Housing, Population Composition, and Offending
Mean house-
hold size.
3.75
3.5
-
3.25
3
2.75
--._-
2.5
._-..
2.
2.25
---
1.75
1.5
_ _ a
1.25
0
2 4 6 8 10 12 14 16
Per cent receiving public welfare
FIGURE 7.6. Scattergram of area mean household size and percentage receiving
public welfare assistance (Stockholm, 1982) (N = 123). Several observations (dots)
are hidden.
Mean house-
hold size.
3.75
3.5
3.25
3
2.75
-
2.5
.- -. - -
2.25
2
1.75
1.5
1.25
0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 27.5
Per cent blue collar workers
FIGURE 7.7. Scattergram of area mean household size and percentage of blue-collar
workers (Stockholm, 1982) (N = 123). Several observations (dots) are hidden.
Familism, SES, Social Problem Households, and Offending 159
Per cent
receiving
public welfare
16
14
. ,.... ..,
12
10
.. . ..-... /:
8
." .
6
.... "'....,.---- ..
............
4
.~..
2 .:" -!.... , .
q~~~~~~~~~~-r~-T~~~-,~-,~~r-r-,
o 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 27.5
Per cent blue collar workers
FIGURE 7.8. Scattergram of percentage in area receiving public welfare assistance
and percentage of blue-collar workers (Stockholm 1982) (N = 123). Several
observations (dots) are hidden.
Zero-Order Correlations
The area total offender rate tends to be rather strongly positively corre-
lated to the proportion of blue-collar workers (SES) and the proportion of
160 7. Housing, Population Composition, and Offending
TABLE 7.13. Zero-order correlations: familism, SES, and social problems with
offending variables (Stockholm, 1982).
Serious
Total Chronic Youth Violent Vandalism property
offenders offenders offenders offenders offenders offenders
Note. Coefficients multiplied by 100. No age-standardized offender rate for youths. N = 121-
123 depending on offender series involved.
Significant at 5% level. Significant at 1% level.
12. The reader is again reminded, that the youth and vandalism offender series are
those, of the included series, that may be the most affected by chance variation, and
the results for these two variables should be interpreted with special care.
Familism, SES, Social Problem Households, and Offending 161
Note. Coefficients multiplied by 100. No age-standardized offender rate for youths. N = 121-
123 depending on offender series involved.
* Significant at 5% level. ** Significant at 1% level.
Multiple Regression
The multiple regression of the joint influence of the three independent
variables on the offender rates shows that a fairly high proportion of the
area offender rates can be explained by area population composition varia-
tions. One difference between using the original and the age-standardized
offender variables is that in the latter, compared to the first case, the pro-
portion of variance explained by the model decreases by 4 to 7% depending
on the offender series concerned (see Table 7.14). The exception is the
subseries for vandalism offenders, where the model is nonsignificant when
using the original offender rate but reaches significance when using the
age-standardized offender rate. However, using age-standardized offender
rates not only means a decrease in the variance explained (where this de-
crease can be attributed to the control for the influence of area age struc-
ture) but also a significant change in the pattern of relative importance
of the independent variables (as determined by the standardized partial
regression coefficients). Although the social problems variable has the
strongest impact of the three on the total offender rate, using the original
rate, it has the lowest impact of the three and almost loses its significance,
when using the age-standardized rate. And as mentioned above, this de-
162 7. Housing, Population Composition, and Offending
crease has to do with the fairly strong relationship between area age struc-
ture and social problems. Looking at the subseries for chronic and serious
property offenders, the area SES is the most important independent vari-
able regardless of whether the original or the age-standardized offender
rate is used as the dependent variable, but when using the age-standardized
offender rate, the social problems variable loses its significance. For the
subseries of violent offenders, just as for total offenders, the social prob-
lems variable is the most important predictor of the three, but unlike
for total offenders, it is still the most important when using the age-
standardized series. The findings thus suggest that area SES is more impor-
tant than area social problems for the area variations in the rates of chronic
and serious property offenders, whereas area social problems is more im-
portant than area SES for the rate of violent offenders. In all instances,
familism is negatively correlated to offender rates.
icant. The exception is the violent offender series, which shows a strong
positive relationship to familism in the group of moderate to high problem
wards. All in all, the findings show that familism has a strong negative
relation to offender rates in low social problem wards but no relationship to
offender rate (except for a positive relationship for violent offenders) in
moderate to high social problem wards.
-....)
:r::
o
c
TABLE 7.15. Range, mean, and median total offender rates for groups of wards combined by familism, SES, and social problem level of '"
area; original and age-standardized rates (Stockholm, 1982). JJ"
Original total offender rates Age-standardized total offender rates
d'
'0
C
;;;l
SES High/medium Low High/medium Low ....
o
::s
Familism High Medium/low High Medium/low High Medium/low High Medium/low n
o
8
Social problems: '0
o
High g1.
....
Range 8.1-29.0 1.8-28.5 20.1-37.7 14.5-31.9 -20.1-5.5 -25.9-11.6 -6.2-9.8 -8.0-12.0 o
::s
Mean 15.9 1404 28.0 23.6 -9.1 -804 0.5 2.2 po
Median 13.2 12.2 26.5 24.2 -10.9 -7.9 -2.7 3.7 ::s
0..
N 4 9 5 13 4 9 5 13
Medium/low
o
~
Range 0.0-21.5 0.0-32.6 (lOA) 4.1-29.6 -26.3-0.5 -24.5-16.0 ( -12.9) -13.0-11.1 ::s
0..
Mean 4.9 11.4 (lOA) 1704 -1604 -7.9 ( -12.9) -1.1 i:i"
-1.6 aQ
Median 4.0 9.7 (lOA) 18.2 -16.5 -lOA ( -12.9)
N 21 56 1 12 21 56 1 12
ly, though not entirely, consistent with the results reported by Polk"
(p.77).
Note. N = 113 when mobility is correlated to offender rates, else N = 121-123, depending on
offender series involved. Coefficients multiplied by 100.
*Significant at the 5% level. **Significant at the 1% level.
for all three measures when using the original rates, whereas when using
the age-standarized rates, the population heterogeneity measure loses its
significance. Of the subseries of offender categories, it is only the violent
offender series that is strongly correlated to all three social disorganization
measures (see Table 7.18).
It is not unproblematic to carry out a multiple regression analysis adding
170 7. Housing, Population Composition, and Offending
the three new social disorganization variables to the three old population
composition variables because the three new variables are intercorrelated
and also all correlated to the social problem variable of the old set of vari-
ables. Gordon (1967) has, in his enlightening discussion of the problems of
multiple regression analysis, with particular reference to ecological studies,
pointed out that
As redundant independent variables are succesively introduced into a regression
problem, their common predictive value gets averaged, in a weighted manner, over
all of their regression coefficients. As a result, all of their regression coefficients
decline in absolute value. At the same time, the multiple correlation increases only
a trivial amount with each new variable, reflecting the fact that little new informa-
tion is being added. (p. 934)
And this is exactly what happens when the three social disorganization
variables are added to the earlier model: The Multiple R2 x 100 increases
by a few percent, but the contribution from each of the three disorganiza-
tionyariables is nonsignificant (as determined by their standardized partial
regression coefficient). The corresponding analyses as reported in Table
7.14 above were run adding the three social disorganization variables.
However, when these analyses were conducted induding only the three
social disorganization variables as independent variables, they all reached
significance for most offender series, showing, so to speak, that the social
problem variable picks up much of the disorganization influence on
offending.
This holds true for the total offender series and for all subseries of offen-
Social problems:
High
% foreigners 10.9 7.3 12.2 6.9
% mobility 2.2 2.1 2.6 2.2
% single parents 24.0 29.8 30.3 30.4
N 4 9 5 13
Mediumllow
% foreigners 3.5 5.4 9.0 5.1
% mobility 1.5 2.2 2.1 2.1
% single parents 7.0 27.4 9.4 24.0
N 21 56 12
Note. Highest and lowest value for each variable in bold type.
Residents moving in or out of residences.
Different Types of Housing 171
TABLE 7.21. Median ward rates of owner-occupied, rented private, and nonprofit
housing (number of areas dominated by type of housing in question in brackets) by
combinations of population composition variables (Stockholm, 1982).
SES High/medium Low
Social problems:
High
Owner-occupied house 18.6 0.1 1.4 2.6
Rented private 10.5 (1) 58.5 (5) 12.3 195 (1)
Nonprofit 38.4 (2) 28.9 (3) 65.0 (5) 66.6 (9)
N 4 10 5 13
Mediumllow
Owner-occupied house 79.2 (20) 0.1 (6) 54.4 (1) 3.6 (2)
Rented private 9.7 68.0 (37) 25.6 29.3 (3)
Nonprofit 0.0 9.2 (2) 4.5 41.7 (3)
N 21 56 1 12
This is further illustrated in Table 7.21, in which it is shown that the high
social problem-low SES areas are heavily dominated by nonprofit housing
areas, regardless of whether they are high or low in familism (14 out of the
24 nonprofit housing areas belong here). Also, the low social problem,
high to moderate SES, and high familism areas are almost exclusively
owner-occupied housing areas (20 of the 29 owner-occupied housing areas
belong here). The low social problem, high to moderate SES areas that are
low to moderate in familism are to a high degree areas dominated by
rented private housing (37 out of 47 rented private housing areas belong
here); moreover, most inner-city areas belong here (29 out of the 35 inner-
city areas).
In this section, I shall explore whether the relationship between area
population composition and offending in Stockholm in 1982 differs among
areas dominated by different types of housing.
16. Balwin and Bottom's (1976) analyses focused on the adult male offender rate,
although other offender category series were considered.
17. That is not to say that detailed studies of any differential patterns of causes of
offending in different types of housing is uninteresting; on the contrary.
174 7. Housing, Population Composition, and Offending
Owner-occupied
house areas
Familism (OR) -22 -23 -27 -25 21 -20
ASOR -45 -32 -40 -14 -33
(low) SES (OR) 58 32 44 51 -07 42
ASOR 52 19 46 -10 39
Soc. Prob. (OR) 47 58 36 46 27 20
ASOR 38 44 41 28 14
N 29 29 29 29 29 29
Rented pivate areas
Familism (OR) -13 -11 10 00 -17 -06
ASOR -28 -13 -11 -33 -13
(low) SES (OR) 31 16 -14 20 -19 37
ASOR 32 17 24 -15 36
Soc. Prob. (OR) 30 06 07 49 -18 20
ASOR 02 -08 32 -29 00
N 45 46 47 47 46 46
Nonprofit areas
Familism (OR) 30 -11 -11 50 28 03
ASOR -13 -28 28 -08 -26
(low) SES (OR) 49 60 -10 19 02 52
ASOR 43 49 13 -13 43
Soc. Prob. (OR) 47 -03 -01 59 40 15
ASOR 02 -26 34 -01 -16
N 24 24 24 24 24 24
Mixed areas
Familism (OR) -35 -35 -18 -23 -13 -33
ASOR -44 -36 -34 -34 -40
(low) SES (OR) 33 26 26 25 22 27
ASOR 28 21 23 29 23
Soc. Prob. (OR) 30 17 23 47 02 19
ASOR 02 -01 23 -24 -02
N 20 20 19 20 20 20
. . ... . .
10
7.5
5
2.5 5
O~----~~--~--~~~~
2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 O.~~~~--~~~~ __~
1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
Non-Profit Mixed Areas
40 35
35
30
30
25 25
20 20
15 15
10 10
5 5
O~~--~~~~--~~--~
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 0
1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4
...
5 10
Or---------~----------- 5
-5 Or--------------------
-5
-10
-10
-15
-15
-20
-20
-25 -25
-302~~2~2~~~~~----~--~ -30,~~~~--~------~--~
. 2.4 2.6 2.8 3 3.2 3.4 3.6 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
-5 -5
-10
-10
-15
-15 -20~~~--~~--~~~~~
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8
1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4
FIGURE 7.9. Scattergrams and regression lines of mean household size (horizontal)
and total offender rate (vertical) for groups of areas by dominating type of housing;
original and age-standardized offender rates (Stockholm, 1982).
176 7. Housing, Population Composition, and Offending
12.5
10 15
7.5 10
5
2.5 5
o0......~25::-+-'.":-5--='.7'="5-,...'":,'""'.2::!5~1-r:.5:-':1:C.7::5:-2"C":2:O_2::5'='2.5 OHh~--~~~~ __~~__~~
o 2 3 4 5 6 7 B
..
5 10
O~------J------------- 5
-5 Or-----~----------------
-5
-10r---~~.~.~~~.~:~-------&------
-15 _
-20
-25 -25
-30 0 .25 .5 .75 1.251.51.75 2 2.252.5 -300~~--~2--~3---4---5--~6---7--~B
'5 20
10 15
5
": 10
5 eo
.
O~======~==========
-5 o _~t=======~~======
-10
-10 -15
-152~~4--~6--~B---1~0---1~2---14--~16 -20~~--~-c--c-------__~
o 234 5 6 7 B
FIGURE 7.11. Scattergrams and regression lines for percentage receiving public wel-
fare (horizontal) and total offender rate (vertical) for groups of areas by dominating
type of housing; original and age-standardized rates (Stockholm, 1982).
178 7. Housing, Population Composition, and Offending
Summary of Findings
Focusing on the total offender rate in Stockholm in 1982, the major
findings of this chapter may be summarized as follows:
1. The total area offender rate shows significant variability and quite a high
degree of concentration to a smaller number of wards. The high total
offender rate wards are generally located in the outskirts of the city
(suburban areas), and most of them are postwar residential develop-
ments (see Chap. 6) dominated by nonprofit housing, although some
high offender rate wards were found in the southern part of the inner
city (the part with the highest percentage of nonprofit housing in the
inner city).
2. The area age-structure variations were found to be no major cause of
the area offender rate variation.
3. Both type of housing and composition of population explains, respec-
tively, quite a high proportion of the total area offender rate variations,
just below 50% in both cases (just below 40% for the population com-
position variables when using age-standardized offender rates).
4. The offender rate showed a clear association to three different measures
(i.e., population heterogeneity, population mobility, and family dis-
ruption) thought to measure aspects of area social disorganization
(although one of the variables-heterogeneity-lost its significance
when using age-standardized offender rates).
Considering the different subseries for offender categories, the series for
youth and vandalism offenders was not very well explained by either hous-
ing or population composition variables, but that is likely to be due to the
fact that these two variables are highly affected by chance variation.
Moreover, as regards the youth offender series, most crimes included in
the present study are of a more serious nature, and many less aggravated
typical youth crimes have not been studied. It is worth noting, though, that
the findings from the cross-table analysis carried out for the Stockholm
1982 area data, and especially the analysis of the individual-level data for
the childhood and youth period from a 1953 Greater Stockholm cohort
(Project Metropolitan), both showed a pattern of offending by housing
equivalent to that for offenders in general.
As for the other sub series of offender categories, the series of chronic
and serious property offenders were highly correlated, and the offenders in
these two categories overlap to a high degree. Therefore, a comparison of
the remaining series can be restricted to the series of serious property and
violent offenders, in which the following significant differences were found:
1. Of the three different dimensions of population composition studied
(i.e., familism, SES, and social problem households), low SES was the
most important predicator for area variations in serious property offen-
ders, whereas social problem households was the most important pre-
An Integrated Model 179
An Integrated Model
In this final section of this chapter, I shall attempt to formulate a simple,
tentative model of the causal ordering of housing, population composition,
and area offender rates and, somewhat, to discuss the links among these
three.
18. It is, of course, always a question of interpretation, but some research designs
are more apt to answer some questions than others. For instance, Kornhauser
(1978), in discussing this problem with reference to the Shaw & McKay research in
Chicago, points to the value of longitudinal studies when trying to separate indi-
vidual and area contextual effects: "The passage of time itself supplies control on
individual attributes, for during this period community populations have changed
or been replaced, especially in slums. Chicago's slum areas still rank fist in rate or
delinquency, no matter what ethnic or racial origin of their occupants" (pp. 115-
116).
180 7. Housing, Population Composition, and Offending
HOUSING POPULATION
COMPOSITION 1--. ...
_~
FIGURE 7.12. Housing, population composition, and offender rate: a simple area-
level model.
about how the design of residential buildings and areas may influence
offending and sometimes even residents' motivation to offend. Generally,
it seems that the hypothesis is that the design of the residential areas and
influences offending in an area through the possibility to exert social con-
trol in the area. The design of the residential area and its buildings is be-
lieved to influence social control possibilities, for instance, by its capability
to make clear what are private and nonprivate areas and to make it possi-
ble for the residents (and others) to surveil what is happening in the area.
A Path Analysis
The simple model of influence suggested above was tested for the Stock-
holm 1982 Crime Survey data, with three variables measuring the percent-
age of each type of housing in an area and the three major population
composition variables (for definitions, see section on population composi-
tion and .offending above) measuring area population composition (where
it is a matter of interpretation, as discussed above, whether one is willing to
interpret the social problem household variables as a measure of social
disorganization or not).
As regards the internal causal ordering between the three population
composition variables, a choice had to be made. One alternative was to
view all three as causally unrelated dimensions of population composition,
but because we know from the previous analysis that area SES and social
problem households are rather strongly related, and that familism is re-
lated to area social problem households in areas with moderate to high
levels of social problem households, it was decided to view the area level of
familism and SES as causally prior to that of social problem households,
especially in light of the possible interpretation of the latter as an area
contextual effect (i.e., as measuring area social disorganization). It was
judged that if there is any causal direction among these variables at the
area level, it goes from familism and SES to Social problems (disorganiza-
tion), not in the other direction. No causal relationship was assumed to
exist at the area level between familism and SES.
The model was tested for the series of total offenders and the subseries
of serious property offenders and violent offenders. The models were run
separately with original and age-standardized offender rates. The full
model is shown in Figure 7.13 for total offenders when using the original
offender rate. For all other models, only the influence of the three popula-
tion composition variables on the offending variable is shown because all
other parameters will be the same as the ones shown in Figure 7.13. Of the
parameters (standardized maximum likelihood estimates) in the model, all
are significant, with the exception of the influence from nonprofit housing
on familism and on SES.
The findings show that the housing of the area explains almost two thirds
of the area variation in familism, whereas its explanatory power on area
182 7. Housing, Population Composition, and Offending
25 52
-39
R2 = 0.94
Chi Square = 19.4 (df=4)
I Non-
prof i t
GFI = 0.96
RMSR = 0.025
i
57
FIGURE 7.13. A path model of housing, population composition, and total offender
rates (Stockholm, 1982).
SES is quite high, but much lower (just above 40%). Considering the pat-
tern of influence from the different types of housing it may be summarized
that high familism goes strongly together with areas dominated by owner-
occupied houses, whereas the relationship to familism for areas dominated
by dwellings rented privately tends to be weakly negative, and for non-
profit areas nonsignificant (including both areas of low and high familism).
As for area SES, the findings show that it tends to be high in owner-
occupied housing areas and in areas with rented private dwellings. In this
connection it should be recalled that most areas dominated by private rent-
ing of dwellings belong to the inner city of Stockholm-almost all areas of
the inner city are dominated by private renting, and many of these areas
also have comparatively high proportions of tenants'-association-owned
flats, although very few areas in Stockholm are dominated by tenants'-
association-owned flats (see Chap. 6).
The area level of social problem households (disorganization) is to a very
high degree (two thirds) explained by the direct and indirect effects of type
of housing in the area and the direct effects of area familism and SES. The
strongest influences come from familism (positive), owner-occupied houses
(negative), and nonprofit housing (positive). A simple way to summarize
this is to say that areas high in social problem households (disorganization)
An Integrated Model 183
TABLE 7.23. Path coefficients (standardized); R2, Chi square, goodness of fit index
(GFI), and root mean square residuals (RMSR) for model; only coefficients shown
for the influence of familism, low SES, and social problems (disorganization) on
original, respective, age-standardized offender rate (Stockholm, 1982).a
Proportion of
offender rate
Path variation
coefficients explained R2 Chi square OFI RMSR
Total offender rate
Familism (OR) -21
(low) SES (OR) 30
Soc. Prob. (OR) 45 48 0.94 19.4 (df=4) 0.96 0.025
Familism (ASOR) -34
(low) SES (ASOR) 40
Soc. Prob. (ASOR) 19 41 0.94 27.9 (df=4) 0.95 0.042
Violent offender rate
Familism' (OR) -17
(low) SES (OR) 15
Soc. Prob. (OR) 63 55 0.94 4.9 (df=4) 0.99 0.013
Familism (AS OR) -26
(low) SES (ASOR) 23
Soc. Prob. (ASOR) 49 49 0.94 8.3 (df=4) 0.98 0.019
Serious property
offender rate
Familism (OR) -15
(low) SES (OR) 40
Soc. Prob. (OR) 17 29 0.94 12.8 (df=4) 0.97 0.033
a All other path coefficients the same as the ones shown in Figure 7.12.
* Nonsignificant (T-value = -0.42).
Note. All path coefficients multiplied by 100.
are areas dominated by nonprofit housing that also are high in familism,
whereas areas low in social problem households (disorganization) are areas
dominated by owner-occupied houses. This is fully in accordance with the
findings presented earlier from the separate analyses of the relationship of
housing and offending, and population composition and offending.
Finally, turning to the influence of housing and population composition
on area offender rate (see Table 7.23), the findings may be summarized in
the following way: In general, all models showed a good fit to data-most
for the model of violent offender rate variations and least for the model of
serious property offender rates. For all models, the major possible im-
184 7. Housing, Population Composition, and Offending
provement of the fit of the modeP9 was to free the influence of nonprofit
housing on the offender rate. In contrast to the other two housing vari-
ables, the influence of nonprofit housing on the area offender rate was
highly significant if freed.
The direct influence of the three population composition variables was,
as expected, the same as in the separate analyses of their relationship to the
offender rate. The same difference as the noted earlier between the sub-
series for violent and serious property offenders appeared also in the path
analysis, namely, that area SES was most important for the area serious
property offender rate, whereas the area social problem household (dis-
organization) level was most important for the area violent offender rate.
However, the path analysis also shows that familism has an indirect in-
fluence on the offender rate, especially so for violent offenders, that goes
over the area social problem households (disorganization). Perhaps one of
the major advantages of carrying out the path analysis is to causally order
th~ influence of housing and population composition on offender rates. It
has been shown that most of the influence of housing on offender rates can
be fit to a model where this influence mainly goes over the effects on
population composition by housing in the area. This draws the attention to
the importance of studying how different social groups are allocated to
different types of housing and the effects of such allocation on the area
offender rate variations-be it predominantly individual, contextual, or
both.
19. As judged by the modification indices available in the LISREL computer pro-
gram that was used for running the path models presented here.
8
Routine Activities, Area Crime, and
Victimization Patterns
The land use in the city varies significantly among its pans. Some areas are
predominantly residential, others predominantly nonresidential (i.e.,
industry, .offices , commerce, entertainments, etc.), and yet others mixed.
The patterns of city land use appear to have some universal similarities as
well as unique characteristics depending on country and city in question. I
shall not go into the reasons for the specific development of the urban
geography and its local variations (see, for instance, Harris & Ullman,
1974; Johnston, 1984). It is sufficient for the purpose of the present study to
conclude that land use varies significantly among parts of a city. The major
patterns of land use in Stockholm have been described in Chapter 6.
The differential land use among areas of a city means that the occurrence
of different kinds of human activities (home life, work, leisure, entertain-
ment, etc.) varies among different parts of the city.! The activity dif-
ferentiation in the city also generates patterns of movement in the city,
which are influenced by street networks and the organization of public
transport (work trips, trips to leisure areas, etc.). Because the occurrence
of different kinds of human activities varies throughout the week and day,
the spatial pattern of human activities and the related movements in urban
space will also vary throughout the week and the day. So, depending on the
land use of an area there will be a specific pattern of activities and move-
ments occurring in that area at different times.
The differential land use in the city is probably the key to the under-
standing of the city's crime patterns. It has been claimed that "since illegal
activities must feed upon other activities, the spatial and temporal struc-
ture of routine legal activities should play an important role in determining
the location, type, and quantity of illegal acts occurring in a given society
1. The concepts of human activities and routine activities will be used interchange-
ably in this chapter. Routine activities have been defined as "any recurrent and
prevalent activities which provide for basic population and individual needs"
(Cohen & Felson, 1979, p. 593).
185
186 8. Routine Activities, Area Crime, and Victimization Patterns
2. It has, however, been argued in later works by Felson (1987, p. 912) that the
routine activity approach can be extended to also include other types of crimes.
Routine Activities, Area Crime, and Victimization Patterns 187
It is often assumed, and some research also suggests, that offenders gener-
ally like to operate in areas familiar to them (e.g., Carter & Hill, 1979).
This also means that targets in areas which offenders are likely to visit and
spend time in, generally speaking, should be more at risk than targets in
other areas.
Depending on what types of crimes are studied, the number of possible
targets is likely to vary among areas of a city. Area differences in the num-
ber of targets may therefore be one influence on the area crime pattern,
although one may argue that there are enough targets for theft, vandalism,
and violence all over the place. Few areas of a city lack things to steal and
vandalize or people to attack. 3 However, if crime were truly randomly
distributed in the city, one would expect that areas with more targets at risk
would have more crime, and hence, one needs to control for targets at risk
in an area before considering more substantive explanations to area crime
variations. 4 This is equivalent to controlling for "population at risk" when
calculating area offender rates (see Chap. 7). I shall return to this problem
below.
However, it is of course possible to find examples of nontrivial hypoth-
eses of the relationship between targets at risk and crime occurrences, for
instance, a "temptation hypothesis" (i.e., other things being equal, more
targets increase the temptations) or a hypothesis that an increasing number
of targets have some necessary (negative) relationship to their level of pro-
tection, other things being equal. A "target at risk" is defined here as any
object (person) that by the definition of the crime in question can be the
target of the crime.
Targets may for various reasons be more or less attractive to a potential
offender, and more attractive targets may be concentrated to some rather
than other areas, which may influence the area crime patterns. Cohen and
Felson (1979) claim that what makes a target suitable is related to its value,
physical visibility, accessibility, and inertia (p. 591). In discussing the con-
cept of attractive (or suitable) targets, it may be helpful to differentiate
between the crimes along two dimension; instrumental versus expressive,4
and interactional versus noninteractional (see Table 8.1). The distinction
between the offender and the target (victim) is generally clear in in-
strumental crimes (e.g., theft, robbery, and rape) regardless of whether
they are interactional or noninteractional. In interactional crimes that are
expressive (e.g., assault) the victim-offender distinction may in quite a few
cases be rather unclear; for instance, many assaults may have the character
of a reciprocal fight rather than a one-way attack (Wikstrom, 1980). In
3. However, in the case of analysis of certain very specific crime types (such as
bank robberies) the target distribution is often likely to be very important simply
because many areas may totally lack targets for the crime in question.
4. As discussed in Chapter 4, this distinction is not always as clear-cut as it may
seem, but as a crude division between categories of crime it may do.
188 8. Routine Activities, Area Crime, and Victimization Patterns
TABLE 8.1. Included acts of crime classified by the dimensions of instrumental ver-
sus expressive and interactional versus noninteractional acts.
Instrumental
YES NO
Interactional
YES Robbery Assault
Rape
NO Burglary Vandalism
Theft of and from cars
crimes that are neither instrumental nor interactional (e.g., vandalism) the
offender-target distinction is generally clear, but in many cases it is difficult
to determine what may constitute a suitable target. Discussion about suit-
able targets in reference to vandalism mostly concerns more or less easy
materials to vandalize.
Building upon the findings from the analysis of the social contexts of
crime in Chapter 4, it will be suggested that although the concept of suitable
targets is generally straightforward for most instrumental crimes, and
generally more so for noninteractional than interactional instrumental
crimes, it is much more problematic when considering expressive crimes.
In the latter cases, there seldom appears to be any more elaborated choice
of target; it is a question of how situations are created that mayor may not
evolve into violence and vandalism rather than a question of a potential
offender taking advantage of a good environmental opportunity. Although
a characteristic of most instrumental crimes is that there is at least a minim-
al search behavior on the part of the offender, this normally appears not to
be the case for expressive crimes.
A special aspect of target vulnerability is the guardianship or lack of
guardianship by so-called capable guardians. A capable guardian may be
any individual who is present and who may be viewed by a potential offen-
der as capable and willing to intervene if a crime were to be committed,
where intervention refers not only to actual physical intervention intended
to stop a crime but also (for civilians) to such things as calling for the police
or being willing to be a witness. Although the police may be the most
obvious capable guardians, it is fairly well-established that police officers
are predominantly reactive rather than proactive (see, e.g., Reiss, 1971;
Knutsson, 1984). Felson (1986) claims that "a capable guardian is seldom a
policeman, more likely a housewife, brother, friend, or passerby" (p. 121).
The guardianship may be stronger in some areas than in others and
stronger at some times of the week and day than at others. Therefore, area
variations in crime may be influenced by area variations in guardianship. It
is predominantly when thinking about instrumental crimes that the concept
of guardianship has its most relevance. For instance, in many assaultive
Routine Activities, Area Crime, and Victimization Patterns 189
crimes, it does not appear that the participants are normally very inhibited,
for instance, by the presence of others (i.e., capable guardians) as they
may be in cases of theft, robbery, and rape (see Chapter 4). The same
appears largely to hold true for the kind of vandalism included in this
study, which is predominantly expressive in nature.
The analysis in Chapter 4 showed that it is important to differentiate
between crimes of violence and vandalism occurring in private and those
occurring in public. The distinction between private and public corre-
sponds to an important distinction among types of routine activities, that is,
those that take place in the home (residential) and those that take place
outside the home (nonresidential). In this chapter, the area distribution of
violence and vandalism in private and in public will be studied separately.
Because crimes in private are almost always committed in residences, it is
the variations among residential areas that are in focus. It will be suggested
here that it is the area residential differentiation of the population that is
crucial for the understanding of area variations of violence and vandalism
in private. Crimes of violence and vandalism in private that are reported to
the police tend to be at the more serious end and generally involve socially
marginalized persons, their spouses, and their mates. It will be hypothe-
sized, and tested in this chapter, that the area crime patterns for violence
and vandalism in private can predominantly be explained by the same
model as that used to explain the area variations in the offender rate (see
Chap. 7).
As for instrumental crimes (excluding rapes and robberies occurring in
private), the city area crime pattern may be hypothesized to be influenced
by area variations in good environmental opportunities (i.e., the amount of
suitable targets with low guardianship) and by the presence of potential
offenders who may take these good opportunities. 5 As for the expressive
crimes occurring in public, it will be suggested here that rather than good
environmental opportunities, it is the amount of friction in meetings be-
tween people that mayor may not generate situations leading to violence
and vandalism that is crucial for the understanding of city area crime pat-
terns for violence and vandalism in public. Although friction in metetings
may seem the most obvious link to violence in public, it will be suggested
on the basis of the findings in Chapter 4 that much of the vandalism that
takes place in public (especially the more serious cases6 ) occurs in the same
kind of circumstances as assaults in public; that is, they are to a large extent
Assault 81 90 83 54
Violence to public official 92 70 58
Robbery 81 62
Molestation 68
Note. All correlations significant at the 0.001 level or better. Coefficients multiplied by 100.
(1) For one reason or another, crimes frequently are not reported at all; (2) there
may be differential reporting of crime from one area to another, which may distort
in varying degrees and directions the derived crime rates for different parts of a
city; (3) there may be differential reporting of crime according to the type or nature
of the crime; (4) sometimes the police, in order to make a good showing or to
protect the name of a community, may suppress or alter official crime records; (5)
the policy of the police in making arrests, as well as the assignment of greater
proportions of officers to certain areas, may bias recorded statistics" (p. 675).
The assumption that will be made in this study, however, is that it is not
likely that area crime rates based on police-recorded criminality will gener-
ally give a biased picture of the pattern of area crime rate variations.
Moreover, it will be further assumed that any bias that will follow from
comparisons of area rates of police-recorded crimes is because area differ-
ences in crime are under- rather than overestimated. This is so, it is argued
here, because it is more likely that people in high crime areas have a higher
threshold for reporting than people in low crime areas (especially as re-
gards less serious crimes), and that the police (if there is any significant
discretion) are less likely in high crime areas than in low crime areas to
record less serious offenses. At least, it appears less likely that any bias
would take the form that more of the crimes actually committed would be
reported in high rather than in low crime areas and that the police would be
more inclined to record crimes in high rather than in low crime areas.
However, the support for these assumptions cannot be claimed to be
dramatic because there are few studies, to this author's knowledge, that
have dealt with aspects of this problem, although the ones that have are
largely in line with the assumptions stated.
Herbert (1978, p. 247) shows, in a comparison of three high and three
low youth offender rate areas in Cardiff, Wales, that residents in the high
youth offender rate areas were less likely to state that minor offenses
should be reported. In the case of serious offenses, there were small differ-
ences in the stated propensity to report between the high and low youth
offender rate areas. Mawby (1979), in a study of nine areas in Sheffield,
England, draws among others, the following conclusion:
The Crime Area Distribution 193
The finding that the discretionary role of the police as regards the discovery and
detection of offences is considerably muted, allied to the apparent willingness of
residents of high crime areas to call the police, seems to question widespread
assumptions that police polices create area differences in crime rate. It is evident
that the recorded information shows no indication of area differences being radical-
ly altered due to different actions of the police (or indeed the public) in different
areas. (p. 247)
In this context it should be mentioned that eight of the nine areas Mawby
studied were working-class areas, although both high and low crime rate
areas were included.
Swedish studies of the types of crimes included here show that only a
very low percentage of the recorded crimes have been detected by the
police themselves; most recorded crimes have been reported by victims
and nonpolice witnessess (e.g., Wikstrom, 1985, p.37).
Smith (1986) finds in a study of 60 neighborhoods in three large U.S.
cities that:
When encounter-level variables such as the type of problem and the race and sex of
victims are controlled for, police are less likely to file reports of incidents in higher-
crime neighborhoods. A threshold effect may operate here, in which offenses must
reach a higher level of seriousness in high-crime neighborhoods before police will
report the incident. (p. 339)
In this context it should be pointed out that one has to allow for the possi-
bility that police practices in these matters may show significant variations
among countries and in some countries even among cities.
All in all, the assumptions that will be made in the following analysis of
the Stockholm area crime patterns are that the police-recorded data overall
will give an adequate representation of the pattern of relative variation in
crime rates among the areas and, if there is any major area bias, it is likely
for the kind of crimes studied here to go in the direction of underestimating
the area differences in crime rates.
7. It should be mentioned here that Harries (1981, p. 155) used opportunity and
risk in the meaning reverse as that in the present study.
TABLE 8.3. Crimes per hour at different time intervals of the day (Stockholm, 1982).a
Nonfamily violence
Time-interval Family & vandalism Violence Vandalism Residential Cellar/attic Nonresidential Thefts of and
of the day violence in residences in public in public burglary burglary burglary from cars
categories of crime is difficult to make; moreover, the temporal pattern for vandalism and serious property crime may also be more subject to bias than that ;l>
for violent crimes. The time-interval classification in the table is an attempt to make an classification broadly corresponding to temporal patterns of activities @
I>'
during a day, for instance, 06:00-08:59 is the normal time when people travel to work, 11:00-12:59 are the normal lunch hours, 16:00-17:59 is the normal
time when people travel home from work. Crimes per hour are calculated in the following way: For the time interval 04:00-5:59 there will be 728 hours for a 9-
year and the crimes occuring in this period have been divided with that number of hours. Crimes per hour for the other classes have been calculated by the
'"S.
a
same principle. :=.
aPeak hours in bold type. o
I:j
......
\0
VI
196 8. Routine Activities, Area Crime, and Victimization Patterns
TABLE 8.4. Suggested best denominators and chosen denominators for the calcula-
tion of city area crime rates.
Suggested Chosen
Crime best denominator denominator
ing in an area; the rationale for using this denominator is the assumption
that there is a rather close correspondence between the number of people
working in an area and the number of companies and public institutions. 8
This denominator has the great disadvantage of not being a direct denomi-
nator, so the rates will not show crimes per targets at risk. The same as is
the case for the rates of violence and vandalism in public. However, it is
assumed that the rate calculated will show, at least crudely, the relative
variation in victimization rates for nonresidential burglaries among the
areas.
The obvious best denominator to use for thefts of and from cars is the
number of cars in an area. However, the use of this denominator means the
same problem as for violence and vandalism in public as regards the possi-
ble disturbance from differences in the temporal distribution of targets and
crimes. Moreover, information on the number of cars in different areas was
not available for the present study. Instead, the same strategy as applied in
the case of violence and vandalism in public was used. The hectare was
chosen as the primary denominator, whereas the residuals from regressing
the number of residents and people working in area on number of theft of
and from cars were used as an alternative measure. The rationale for this is
the assumption that there is a fairly strong relationship between the num-
ber of people residing and working in an area and the number of cars in the
area.
The definitions of the eight primary crime series are listed below, fol-
lowed by the result from the multiple regression analysis carried out to get
the alternative measure for violence in public, vandalism in public, and
thefts of and from cars (see Table 8.5). It should be pointed out that only
cases of pure vandalism (see Chapter 4) have been included in the area
series of vandalism.
Family violence. Crimes of violence (homicides, assaults, robberies,
rapes, and molestations) within the family per 1,000 households.
Nonfamily violence and vandalism within the home. Crimes of violence
(homicides, assaults, violence to a public official, rapes, and molesta-
tions) and crimes of vandalism occurring in recidences or their proximity
per 1,000 residences.
Crimes of violence (homicides, assaults, violence to public officials,
rapes, and molestations) occurring in public places (streets, sqaures,
parks, public transport) per hectare.
Crimes of vandalism occurring in public places (streets, squares, parks,
and on public transport) per hectare.
Thefts of and from cars per hectare.
Violence in public
Number of residents -0.003 -0.117 -2.02 0.046
Number of working in area 0.013 0.798 13.81 0.000
Multiple R2 x 100: 61.4
Vandalism in public
Number of residents 0.004 0.383 5.50 0.000
Number of working in area 0.003 0.473 6.80 0.000
Multiple R2 x 100: 44.1
Thefts of and from cars
Number of residents 0.042 0.475 8.70 0.000
Number of working in area 0.033 0.570 10.44 0.000
Multiple R2 x 100: 65.6
nonprofit housing problem areas in the outskirts. In the former case, the
relative rates were significantly lower when using the alternative (residual)
measure than when using the crimes per hectare measure, whereas in the
latter case the direction of the difference was the reverse.
The relative merits and problems of the two alternative measures of
crimes of violence and vandalism in public and crimes of theft of and from
cars will be further discussed in connection with their use in the different
analyses below.
TABLE 8.6. Range, median, percentage of zero-rate wards, and measures of skew-
ness for the crime series.
% wards with
Series Range Median zero rate Skewnessa
Extreme Values
An inspection of extreme values (defined in the same manner as was the
extreme values for the offender residential rates in Chap. 7) revealed that
none of the extreme rates could be attributed to factors that would moti-
vate their removal from the analysis. Therefore all identified extreme
values were included. The most significant extreme values were found for
violence and vandalism in public in one of the two CBD wards. The same
holds true when using the alternative (residual) measure, although the
extreme value was less extreme.
9. In this context it should be mentioned that not all of the studies referred to in
this section have measured the crime rates per residents, or the crimes per residents
only.
202 8. Routine Activities, Area Crime, and Victimization Patterns
6 .6-
4.6 - 6 .0
.3.1 - 4.5
g] 1.6 - 3.0
o 0.0 - 1.5
inc1.
Non-femily crimes
in privele,
per 1.000 residences.
9 .7-
7 .3 - 9.6
4 .9 - 7.2
g] 2.5 - 4.8
o 0.0 - 2.4
inc1.
MAP 8.2. Nonfamily violence and vandalism in residences per 1,000 residences
(Stockholm, 1982).
204 8. Routine Activities, Area Crime, and Victimization Patterns
Violence in public
per hectere;
. 2. 1 -
1 .6 - 2 .0
1. 1 - 1.5
fKl 0.6 - 1.0
D 0.0 - 0 .5
incl.
Vende1tsm in public
per hectere;
. 2. 1-
1.6 - 2 .0
1. 1 - 1.5
Not incl .
Residentiel burgleri es
per 1.000 res ide nces .
41 -
31 - 40
21 - 30
[] 11 - 20
0 0 - 10
incl.
Burglaries in cellars}
attics per 1.000 residen-
ces in multi~torey houses .
41 -
. 3 1 - 40
. 2 1 - 30
EJ 11 - 20
o 0- 10
i ncl.
MAp 8.6. Burglaries in cellars and attics per 1,000 residences in multistory houses
(Stockholm, 1982).
206 8. Routine Activities, Area Crime, and Victimization Patterns
41 -
.31 - 40
21 - 30
[ill 11 - 20
o 0- 10
incl .
MAp 8.7. Nonresidential burglaries per 1,000 people working in area (Stockholm,
1982).
8 -
.6-7
.4-5
r;:;:::]
[:;:j 2 - 3
00-1
incl.
MAP 8.8. Thefts of and from cars per hectare (Stockholm, 1982).
The Area Crime Rate Distributions 207
Violence in public.
Residue.! measure.
237 -
132 - 235
27 - 131
lIT! 78 - 25
0 - 184- -79
j ncl.
Vandelism in public
Residue.! mellSUre.
72 -
38 - 71
4 - 37
8ill -30 - 3
D -65 - -31
incl.
376-
144 - 375
- 88 - 143
[ZI -320 - - 89
D -553 - -321
Not i nc l.
MAp 8.11. Thefts of and from cars; residual measure (Stockholm, 1982).
belong to the upper quartile wards, and the majority of the wards in the
southern part of the inner city belong to the upper quartile wards.
The major difference between the two measures for all three series is
that many of the wards in the northern part of the inner city that belong to
the upper quartile of ward rates when using the primary measure, do not
belong to the upper quartile when using the residual measure. Another
major difference between the two measures is that more outer-city wards
belong to the upper quartile wards when using the residual measure. All in
all, the extreme pattern of the location of high-rate wards to parts of the
inner city for violence and vandalism in public, and thefts of and from cars,
diminishes somewhat when using the alternative (residual) measures.
10. The reader is reminded that most of the thefts of and from cars takes place
when the car is parked in streets (see Chap. 4).
11. Meaning that the same variables were loading high (i.e., 0.40 or higher) on
respective factor.
tv
.....
o
00
TABLE 8.7. Zero-order correlations between the area crime series (Stockholm, 1982) (N = 123). S'
~
o.
~
Family Nonfamily Violence Vandalism Residential Cellars/attics Nonresidential Car o
violence violence in public in public burglary burglary burglary theft (")
>
....
Primary crime series ~:
....
Family violence 100
o
Nonfamily violence
'"
& vandalism 22* 100
>
....
o
~
TABLE 8.8. Factor loadings and communalities; Varimax rotated factors (Stock-
holm, 1982) (N = 123).
Crime series FI FII Fill h2
Primary series
Family violence 7 68 - 1 46
Nonfamily violence and vandalism 70 -28 57
Violence in public 86 - 2 6 75
Vandalism in public 93 9 0 88
Residential burglary - 2 - 5 95 90
Cellar/attic burglary 19 50 25 35
Nonresidential burglaries -15 81 0 69
Thefts of and from cars 86 2 - 4 74
Contribution % 31 23 13
in the inner city. The outer-city high-rate ward pattern for Crimes in
Private and Against Semipublic Space is markedly different from that
for Crimes Against the Private (Residential Burglaries). When using the
alternative (residual) measures for Crimes in Public in the factor analysis,
the inspection of a map plot of the factor scores (not shown) for this factor,
shows that the major difference is that the high-rate wards tend to be less
extremely inner-city concentrated.
The pattern of relationship between the area crime series was also ana-
lyzed separately for the inner-city and the outer-city wards. The zero-
order correlations are shown in Table 8.9. The findings from the factor
analyses carried out separately for the inner city and the outer city (not
shown) are discussed below.
The separate factor analysis for the outer-city wards (not shown), using
the primary crime measures, gave basically the same result as for the whole
city. Three factors with an eigenvalue above 1 were extracted (together
explaining 65% of the variance). The same variables were loading high
(i.e., .40 or higher) on respective factors as in the analysis of the total city,
although one significant difference was that the factor of Crimes in Private
and Crimes against the Semiprivate was stronger than the factor of Crimes
in Public (the former contributing with 36%, and the latter with 15%, to
the explained variance).
The result of the separate factor analysis for the inner-city wards (not
shown) was somewhat different. Four factors were extracted with an eigen-
value above 1 (together explaining 77% of the variance). The basic factor
structures identified for the whole city and the outer city separately were
not fully reproduced for the inner city. Only the factor of Crimes in Public,
which was the strongest factor (contributing with 31% to the explained
TABLE 8.9. Zero-order correlations between the area crime series, separately for the inner city (N = 35) and the outer city (N = 88); N
outer-city correlations given below inner-city correlations (Stockholm, 1982). .....
N
Note. Coefficients multiplied by 100. Significant correlations in bold type, A higher value on the correlation is needed for the inner-city than for the
outer-city series to reach significance because the observations are fewer in the former case,
*Significant at 5% level.
., Significant at 1% level.
Crime and Distance 213
variance), was the same. However, because the inner-city wards are rather
few (N = 35), one should interpret the findings for the inner city with some
care-there is room for significant chance variation, although it appears
likely, given the overall environmental difference between inner-city and
outer-city areas, that some difference in the area crime patterns and the
associations between the crime series would show in a separate analysis of
the inner city and the outer city.
Introducing the alternative (residual) measures for the series of crimes in
public (violence, vandalism, thefts of and from cars) resulted in some
changes in the factor structure in the separate analysis of the outer city (not
shown). Three factors were extracted with an eigenvalue above 1 (together
explaining 60% of the variance). The major difference in the factor struc-
ture, compared to when the primary measures of the series of crimes in
public were used, was that thefts of and from cars did not load high (Le.,
over .40) on any of the three factors, and, moreover, nonresidential bur-
glaries loaded high not only on the factor of Crimes in Private and Crimes
Against Semiprivate Space but also on the factor of Crimes in Public.
All the findings from the factor analysis so far presented have been based
on orthogonal rotations (Varimax) because the oblique rotations (Pro-
max), which also were performed, showed very low interfactor correlations.
However, for the separate factor analysis of the outer city, when using the
alternative (residual) measures for Crimes in Public, showed that two of
the factors were moderatley positively intercorrelated (0.27), namely, the
factors of Crimes in Public and Crimes Against the Private (Residential
Burglary). As to using the alternative (residual) measures in the separate
factor analysis of the inner city, the result was close to that achieved when
using the primary measures, although only three factors with an eigenvalue
above 1 were extracted (together explaning 64% of the variance)..
All in all, the area crime series relationship pattern identified for the
whole city appears to be fairly robust, although the separate analysis for
the inner city showed a stronger deviation from the total city pattern than
did the separate analysis for the outer city. Partly one is to expect that the
patterns for the inner city and the outer city should show some differences;
the difference may to some extent be influenced by the fact that the chance
variation may be greater for the inner city taken by its own, given its rather
few numbers of observation. I shall return to the question of explaining the
area city crime pattern later on, but before that, the issue of crime and
distance will be treated in the next section.
Turner, 1969a; Pyle, 1974, pp. 142-180; Capone & Nichols, 1976; Baldwin
& Bottoms, 1976, pp. 78-98; Phillips, 1980; Rhodes & Conly, 1981; Mc-
Iver, 1981; Wikstrom, 1985, pp. 212-230).
In general, it has been reported in Anglo-American studies of crime and
distance that there is a "distance-decay" function-that is, the number of
crimes committed tends to decrease with the distance from the offender's
residence-and that the distance traveled tends to be shorter for violent
than for property crimes. For instance, Phillips (1980) concludes from a
review of the literature that "all researchers have reported finding a
distance-decay relationship" (p. 168), and many studies, including Phil-
lips's own (p. 169), show that, on average, violent crimes are committed
closer to the offender's home than are property crimes (see, e.g., Pyle,
1974, p. 149; Baldwin & Bottoms, 1976, p. 81). Partly this latter finding is
expected, because quite a few violent crimes take place in the offender's
residence (see Chap. 4), which is almost never the case in property crime,
although some U.S. findings show for violent crimes that, even after the
subtraction of family violence cases, the distance between the offender's
residence and the place of crime commisson was short (Porkny, 1965).
It has been pointed out by Turner (1969a) that "if the delinquent lives
close to his offense, there would be little difference between defining a high
delinquency area in terms of offender's residence or offense location" (p.
25). If this were generally true, it would also mean that the same models
used to explain area variations in offender residence rates (see Chap. 7)
would be applicable to the explanation of the area crime rates. Turner
himself draws the conclusion from his Philadelphia study that the high
offender residential and high crime areas are basically the same. However,
this is not a universal finding.
Baldwin and Bottoms (1976, p. 97), commenting upon Turner's findings,
state that offense and offender areas could not be regarded as synonymous
in Sheffield. Swedish studies of, for instance, property crimes in Malmo
(Werner, 1974), and assaults in Gavle (Wikstrom, 1980), also show that
the area distribution of the crimes and the offender residences are far from
identical.
These differences in findings may at least partly be explained by differ-
ences in patterns of housing among different cities. The fact that the high
offender residential areas in many big U.S. cities tend to be located near to
the city center, which generally is the area of the city where most crimes, at
least numerically speaking, take place (see above), means that the differ-
ence in location between high crime and high offender residential areas
tends to be less significant in big U.S. cities than in big British and Swedish
cities, where, at least nowadays (see Chap. 7), high offender residential
areas are less often located close to the city center. Brantingham and
Brantingham (1984) claim that "in cities with a free housing market, crim-
inal areas tend to be located near the central business district. In cities
with zoning and extensive public-housing programs, criminal areas may be
Crime and Distance 215
00
:;0
TABLE 8.lD. Crimes and distance by type of crimes (Stockholm, 1982). o
J:
O.
::I
Location of crime n>
in relation to % Family % Nonfamily % Violence % Vandalism % Residential % Cellar % Nonresidential % Car ;:.-
offender residence violence violence in public in public burglary burglary burglary theft ~
~:
In offender's G'
V>
residence or
proximity" 56.7 51.5 1.7 4.2 0.5 3.3 1.4 1.1 ..,;:.-
n>
po
In offender's
neighborhood b 8.3 13.8 16.7 15.5 18.8 23.4 21.6 17.2 n
::1.
Other place in S
n>
Stockholm 21.7 20.1 42.6 43.7 54.4 60.0 24.5 52.3 po
Offender living ::I
0-
outside Stockholm 13.3 14.6 39.0 36.6 26.3 13.3 24.5 29.4
<:
~.
Median distance" 0.0 0.0 27.0 20.0 25.0 34.0 23.0 30.5
S
No. of included cases 834 1,337 2,177 710 1,360 360 834 2,958 N'
Missing cases 103 269 1,775 3,470 3,944 4,710 9,408 23,205 ~
o
Total cases 937 1,606 3,952 4,180 5,304 5,070 10,242 26,163 ::I
% missing 11.0 16.7 44.9 83.0 74.4 92.9 91.9 88.7 "tJ
~
n>
"Proximity = within 100 meters of the residence. ::I
V>
bWithin 1,000 meters from the offender's residence, excluding cases in offenders residence/proximity.
"In hundreds of meters; distances measured only for offenders residing in Stockholm.
Crime and Distance 217
present study because its scope includes only those crimes occurring in the
city. Finally, it should be mentioned that there is a smaller group of offen-
ders with no fixed address. If they were living with some permanence in an
institution, hostel, or other such place, this address was used as their home
address otherwise, they have been referred to as missing cases in the eco-
logical analysis.
Many crimes lack a known offender, and this proportion is much higher
for vandalism and property crimes than for violent crimes (see again Table
8.10). Insofar as the unknown offenders' crimes are distributed differently
as regards distance traveled compared to those of the known offenders, a
source of bias is introduced. It may, for instance, be the case that offenders
not living in the area where the crime is committed may have a lesser
apprehension risk than the local offenders. Unfortunately, very little is
known about this.
The distance from the offender's home to the place of crime commission
can be qteasured in several different ways. The two most common mea-
sures are (1) the shortest line between the two points (line distance), and
(2) the shortest way by road (wheel distance). Arguments may be raised
against the use of both these measures. I have used line distance for this
study. Wheel distance is a much more time-consuming measure to obtain
and, as Phillips (1980) states, "there is no reason to assume that the short-
est street distance measured from a map has any relation to actual travel
routes" (p. 173). Phillips states further that both of these methods yield
very similar results, although that opinion is not agreed with by all research-
ers. Rhodes and Conly (1981, p. 178) point to such factors as natural
barriers (e.g., rivers) that, according to their view, make wheel distance a
more appropriate measure.
In measuring the distance between the o.ffender's home and the place of
crime commisson for this study, the goal has been to have errors not above
100 meters from the various measured points. Although larger errors are
likely to occur in some instances, there is no reason to believe that these
would in any way be systematic (i.e., correlated with other studied vari-
ables) but rather random. Crimes occurring just outside the residence of
the offender have been referred to as zero distance because, as noted, it
was not judged possible to have a precision finer than hundreds of metres.
?O
::0
0
~
::to
::I
Cl>
TABLE 8.11. Distance between offender residence and place of crime commisson in hundreds of metres; only cases where offender resided >
~
in Stockholm (Stockholm, 1982). ~:
~
Note. Peak percentages in bold type. The maximum possible length of distance traveled for an offender living in Stockholm and committing a crime there is
300, or 3 Swedish miles (1 Swedish mile is, roughly speaking, 6 English miles). However, the shape of the city of Stockholm makes a distance as long as that a Q
rather unlikely event. s
(!)
0-
9-
CJ>
(')
(!)
N
.....
\0
220 8. Routine Activities, Area Crime, and Victimization Patterns
victim's residence. 12 This means that quite a big number of the offenders in
this type of crime are neighbors to their victims.
Second, the findings for the other crime categories-violence and van-
dalism in public and the property crimes-show that there is a distance-
decay pattern with a concentration of places of crime commission to the
area close to the offender's homeY although zero distance for these
crimes, for natural reasons, is uncommon-that is, by definition of the
crime types, there will be no crimes taking place in the offender's resi-
dence, so the zero distance cases all refer to cases taking place just outside
the offender's residence. Roughly speaking, between 20 and 30% of the
crimes of vandalism and violence in public, and the property crimes, occur
within 1 kilometer of the offender's home, including zero distances (see
Table 8.11). This percentage is, of course, lower if those offenders not
residing in Stockholm are included (see again Table 8.10). When excluding
violence and vandalism in private (family violence, and nonfamily violence
and vandalism in residences), there is no clear pattern that violent crimes,
on average, tend to be committed closer to home than property crimes,
especially when taking into consideration that violence in public is the
crime category that has the highest proportion of offenders not residing in
the city (see again Table 8.10).
All in all, the findings show that (with the expection of crimes in private,
which are predominantly local crimes) although a significant pmportion of
the crimes are locally committed (i.e., within 1 kilometer of the offender's
residence), the majority of the crimes occur quite a distance away from the
place were the offender resides. Between 40 and 55% of the crimes, de-
pending on crime type, occur more than 3 kilometers from the offender's
residence (see again Table 8.11), and if the non-Stockholm resident offen-
ders are also included these proportions will, of course, increase.
12. The reader is reminded that in the ecological analysis all cases of nonfamily
violence in residences are included, not only those occurring between acquain-
tances.
13. It should, in this case, be noted, that even though all the distance classes (with
the exception of the 0 distance and the 180 and greater distance classes) have equal
distance intervals, the higher the distance class, the greater is the area of land
covered. For a discussion of this, see Turner (1969a, pp. 14-17).
Crime and Distance 221
analyzed below, whereas in this section I shall concentrate on the area offen-
der rates for local offending (outside the private) and nonlocal offending.
The definitions of these two series are as follows:
Local crimes are those in the offender's neighborhood, that is, within
1 kilometer from the offender's residence, excluding cases occurring in
the offender's residence or its proximity .
Nonlocal crimes are those outside the offender's neighborhood, that is,
more than 1 kilometer from the offender's residence.
One outer-city ward was found to have extreme values both for local and
nonlocal crimes, due to one serial clearance case, and this ward was there-
fore deleted from the analysis.
In Chapter 7, it was demonstrated that the offender residential distribu-
tion showed significant area variation and that this variation to a high de-
gree could be explained by the area variations in population composition
(and possibly also its related social consequences), which in turn was highly
related to area variations in housing. In this section, the question will be
raised as to whether the offender residential distribution affects the area
crime rate: Do high offender residential areas have a higher rate of local
offending than other areas? To answer this question, one cannot just corre-
late the offender residential distribution with that of the crimes, because
such a correlation says nothing about whether or not it is the residents who
commit the crimes in the area. Another-approach was applied here.
The total area offender rate was strongly correlated both to local offend-
ing (r = .69, p = 0.000) and to nonlocal offending (r = 0.75, P = 0.000) by
the residents. The area offender rates of local and nonlocal offending were
also significantly positively correlated to each other (r = .30, P = 0.000,
N = 122). Regressing nonlocal offending on local offending and inspecting
the wards with high residuals revealed that the five wards with the highest
rates of nonlocal offending (rates of 15 or more) all had a very low rate of
local offending-in four out of five cases the rate of local offending was O.
When these five wards were deleted from the computation, the value of the
correlation increased significantly to .52, but this is just to say that there
are some significant outliers in the regression that markedly reduce the
value of the correlation between local and nonlocal offending. The area
distributions of local and non local offending is shown in Maps 8.12 and
8.13.
All in all, the impression is that local and n~~local offending by residents
are distributed in rather the same way. But this does not,necessarily mean
that the offender residential distribution is a good predictor of the area
crime distribution, just that living in a high offender residential area means
that one can expect one's neighbors to commit more crimes in the neigh-
borhood than if one were living in a low offender rate neighborhood. There
are many factors other than the offender residential distribution that may
explain the area crime rate variations.
222 8. Routine Activities, Area Crime, and Victimization Patterns
13 -
10 - 12
7- 9
EJ 4- 6
0 0- 3
inci.
Non-locally committed
crimes per 1.000 inheb.
10 years or older.
21-
16 -20
11 -15
ffiJ 6 -10
0 a -5
i ncl.
Primary series
Family violence 40**
Violence and vandalism in
residences 40**
Violence in public 16
Vandalism in public 28**
Residential burglary -04
Cellar and attic burglary 13
Nonresidential burglary 32**
Thefts of and from cars 29**
Alternative measures
Violence in public 15
Vandalism in public 26**
Thefts of and from cars 25*
a"Local offenders" are deli.ned as those offenders not living more than 1 kilometer from the
place of crime commission.
bNo crimes of this kind occurring in area, or too few to calculate percentage.
Familism -4 -2
LowSES -1 8
Social problem 62** 49**
Multiple R2 x 100 38.1 28.6
ary, although the association was not dramatically high. Against this back-
ground it seemed reasonable to hypothesize that the same factors that
explained the area variations in the offender rate also would explain area
variations in the rate of crimes in private. Because it was shown in Chapter
7 that the influence of housing on offending mainly relates to its effects on
the area population composition, it was decided to restrict the multiple
regression analysis to the influence of the three major population composi-
tion variables of crimes in private. The findings (see Table 8.14) show that
both kinds of violence are significantly related to variations in the area
population composition; however, the only significant predictor of the
three variables is the social problem variable, which may be interpreted
alternately as measuring area social disorganization (as discussed in Chap.
7).
19 and below 11 31
20-29 23 137
30 and above 32 105
a In hundreds of meters.
Violence in public
Familism -4 -19
LowSES 7 4
Social problem 0 44**
Public entertainment 81** -5
Multiple R2 x 100 63.9 27.2
Vandalism in public
Familism -17* -6
LowSES -17* -15
Social problem 26** 65**
Public entertainment 57*' -4
Multiple R2 x 100 51.3 34.0
vandalism in public and, studying the outer city only, to .14 for violence in
public and to 0.25 for vandalism in public, using the alternative measures
on violence and vandalism in public.
As reported earlier in this chapter, the area variations in thefts of and from
cars were strongly correlated to the area variations in violence and vandal-
ism in public (see again Table 8.7). In Chapter 4, it was shown that most
thefts of and from cars occur when cars are parked on the street and, for
those cars belonging to private persons, mostly when they are parked in
proximity to the owner's home (i.e., within 500 meters of their residence).
The map (see again Map 8.8) of the distribution of car thefts showed that
the highest rates were found in the city center of Stockholm or areas near-
by the city center. The distribution for the suburban areas separately (not
shown) ~hows that the rates of thefts of and from cars tends to be highest in
the problem areas.
Assuming that the highest presence of motivated offenders will be in the
problem areas (where offenders are likely to reside) and in the city center
(where they are likely to spend quite a lot of their time), One may hypothe-
size that the area variations in the rate of thefts of and from cars is related
to area variations in the presence of motivated offenders. In this context it
can be noted that the area rates of thefts of and from cars showed a positive
and significant relationship to the area rate of local offending (see again
Table 8.12).
To make a crude test of these assumptions, the same independent vari-
ables as those used for explaning the area variations in the rate of violence
and vandalism in public were used in the analysis of the area rate variations
for crimes of theft of and from cars. However, in this analysis the res-
taurant variable is rather thought of as an indicator of the presence of
motivated offenders rather than a measure of the production of friction in
meetings/situations. The reader is reminded, as reported in Chapter 4, that
Swedish studies of persistent criminals/delinquents show this group to have
a very high participation in public entertainment activities (see also the
description of the CBD area in Chap. 6). For this reason, the use of there-
staurant variable can be justified, at least, as a crude measure of the nonre-
sidential presence of motivated offenders in an area.
The findings (see Table 8.18) give some at least weak support for the fact
that the highest rates of thefts of and from cars are likely to occur in areas
where motivated offenders reside or tend to spend time. Of further interest
in the analysis presented in Table 8.18 is the significant negative association
of the rate of thefts of and from cars to area familism. This is interesting
from a guardianship point of view, because many of the areas high in famil-
ism are single-family house areas, and the guardianship of cars parked in
such areas (often at the ho.use lot or in a garage) may be considered to be
232 8. Routine Activities, Area Crime, and Victimization Patterns
higher than in other areas where cars are are mostly parked in public
spaces. Supporting this view is the rather high negative zero-order correla-
tion between the percentage of single-family houses in an area and the rate
of thefts of and from cars (r = - .46, p = 0.000). Therefore, it may be sug-
gested that the evidence points to the fact that the lowest rates of thefts of
and from cars tend to be found in areas where guardianship as regards cars
parked in the area is highest.
As in the analysis of violence and vandalism in public, using the alter-
native measure of the rate of thefts of and from cars in the analysis (not
shown) significantly reduced the proportion variance explained by the
model, although the pattern of single variables' influence on the crime
rate was the same. However, in the analysis of the total city (Multiple
RZ = .10), only social problems was a significant predictor, and in the
analysis of the outer city (Multiple RZ = .06) no one of the single indepen-
det predictors reached statistical significance at the 5% level. The question
of whether one is more willing to believe in the accuracy of the results using
the alternative or the primary measure is left to the reader to decide,
although it should be noted that the major difference between the use of
the two is the level of variance that is explained by the model.
TABLE 8.19. Victimization rates per 1,000 residents aged 10 or older. Range,
median rates, percentage of wards with zero-rate, and a measure of skewness of
distributions (Stockholm, 1982).
Series Range Median % zero rate Skewness a
Victims of violence
per 1.000 inheb.
10 years or older .
21-
16 - 20
11 - 15
[] 6 - 10
o 0- 5
inc1.
MAP 8.14. Victimizations for violent crime per 1,000 inhabitants aged 10 or older
(Stockholm, 1982).
inheb. 10 years or older .
73 -
55 - 72
37 - 54
181 19 - 36
0 0- 18
incl.
MAp 8.15. Victimizations of serious property crimes per 1,000 inhabitants aged 10
or older (Stockholm, 1982).
The Area Distribution of Victims of Personal Crime 235
TABLE 8.20. Zero-order correlations between area rates of offenders and victims of
violent crimes and serious property crimes (Stockholm, 1982) (N = 122).
Offenders Victims of violent crimes Victims of serious property crimes
Violence 60" 10
Property 30** -11
Familism -21 *.
Low SES 20'
Social problem 47'*
Public entertainment 17*
Multiple R2 x 100 42.9
In this concluding chapter, I shall present and discuss the general picture of
urban crime, criminals, and victims that has emerged from the findings
presented in the earlier chapters. I shall also try to link the findings from
the four different approaches to the study of crime and offending that have
been treated in this book.
237
238 9. Integration and Discussion of Findings
Urbanization
/
!
Greater oppor-
tunities
1
More criminal
events
FIGURE 9.1. The relationship between urbanization and crime: a tentative model.
TABLE 9.1. Characteristics of main types of social orders following Hunter (1985).
Private Parochial Public
reasons, more densely built than others-but is important for social life in
that it is likely to further influence the rate of stranger-to-stranger meetings
and hence contribute to the importance of the social life occurring in the
public domain.
The urban area is characterized by a mixture of commercial, industrial,
administrative, and entertainment activities and, partly related to that, by
a social mix of its population. The degree of heterogeneity in activities, and
in the social background and characteristics of the population, is likely to
show a positive relation to population size, although far from any one-to-
one relationship. Different cities of the same size may vary considerably in
their heterogeneity. Degree of heterogeneity is important for social life
because it influences the degree of meetings and interactions with people of
different social backgrounds and the degree of contact with different kinds
of activities.
Social control is a major causal factor in many theories of crime (e.g.,
Hirschi, 1969; Kornhauser, 1978), and it is also a factor often pointed to as
an explanation for the positive relationship between the degree of urban-
ization and the rate of crime. Increasing urbanization is hypothesized to
weaken social control.
Social control has many forms. Following Kornhauser (1978, p. 74), one
may distinguish among different forms of social control after the dimen-
sions of internal-external and direct-indirect controls. Internal control re-
fers to the conscience (direct) built up in the socialization process and to
rewarding social bonds with others (indirect). External control refers to
general prevention from the risk of sanctions or of losing rewards in others'
command (indirect) and surveillance (direct) .
It seems to be a reasonable hypothesis that there is a relationship be-
tween degree of urbanization and the conditions for social control. Direct
and indirect internal controls are predominantly developed and maintained
in the private and semipublic (parochial) spheres of people's lives, and
social institutions, such as the family, the school, and peer groups, all play
important roles as agents of social control at different stages in a person's
life. It is highly plausible that social institutions of the type mentioned, in
240 9. Integration and Discussion of Findings
their role of fostering children and youth, work under different conditions
in areas of different population size and heterogeneity. One may hypothe-
size that the development of a strong internal control generally is more
easily achieved in smaller, more homogeneous areas than in larger, heter-
ogeneous areas. As for external controls, it is highly plausible that the
control that result from the monitoring and surveillance of other's behavior
is generally speaking a more difficult task in bigger as compared to smaller
settlements. At the same time, it is likely that direct external control plays
a greater role in bigger places because a greater part of the social life in
such places takes place in public and involves strangers.
If weaker social control is one of the two major explanations, for the
relationship between degree of urbanization and rate of crime, greater
opportunities for crime is the other. The size, density, and heterogeneity of
an area is likely to influence the amount of available goods to steal and the
amount of meetings with other people of different social backgrounds that
may develop into conflict and violence. If by the phrase opportunity for
crime is understood the presence of good environmental conditions to com-
mit crime (i.e., a suitable target in the absence of guardianship), the degree
of direct external social control (see above) in an area is likely to be an
important influence on the amount of good environmental opportunities
that a resident in the area is confronted with.
However, as discussed in Chapter 8, one may question the relevance of
direct external social control in influencing the occurrence of expressive
forms of violence and vandalism, because many expressive crimes of vio-
lence involve formal agents of control (police, doormen, etc.) and because
the study of these types of criminal events shows that in many cases the
parties appear to take little notice of whether capable guardians are pres-
ent or not (see Chap. 4).
Together with other motivational forces that may operate in the area
(dependent on structural and cultural characteristics of the society in ques-
tion), the average success of an area in developing among its residents
strong internal controls against breaking the law may be assumed to in-
fluence the rate of motivated offenders in the area (or perhaps more cor-
rectly, acknowledging that offending is a variable-the rate of individuals
with a higher propensity to offend). The rate of motivated offenders in an
area may also be hypothesized to be influenced by the general level of
temptations, related to the degree to which the residents are confronted
with good environmental opportunities for crime.
Perhaps the most important aspect of Cohen and Felson's (1979) work
on routine activities and crime is their emphasis on, for a crime to occur,
the necessary convergence of, minimally, one motivated offender and a
suitable target in the absence of guardianship (although the guardianship
aspect can be questioned for many expressive forms of violence, as already
stated). This implies that a given city structure (i.e., the inner differentia-
tion of the residential population and routine activities)-regardless of the
The Urban Offender 241
into crime. This may also explain the low female rate of offending, assum-
ing that females, generally speaking, are through the socialization process
likely to develop stronger (direct and indirect) intemal controls than males.
The same line of reasoning may hold true for the social class differences
noted in Chapter 3.
It is well-known that there is a significant residential segregation of dif-
ferent social groups in the urban area. (Patterns of segregation may show
some differences between cities and countries.) It is also well-known that
the offender area residential distribution is highly skewed and that the
offenders are differently distributed from the normal population by social
groups (e.g., by social class or family status). Finally, it is well-established
that the offender area distribution shows a strong association to area
population composition (family status, household socioeconomic status,
problem households), which, in turn, is dependent on housing allocation
mechanisms, which may be different in different countries (see Chap. 7).
The question as to what degree the residential segregation of offenders is
due to social group segregation or area contextual effects is a key issue in
studies of the patterns of urban crime and one that still largely remains
unsolved. The main contextual variable pointed to in the literature on
urban crime is the possible influence of residential area social disorganiza-
tion (i.e., community structural difficulties in establishing common values
and solving common problems) on children's and youths' propensity to
offend. Degree of area social disorganization is generally seen as a result of
population heterogeneity and mobility and family disruption. Area social
disorganization means that important social institutions, such as the family
and the school, do not as effectively succeed in fostering strong internal
controls among its children and youth as they do in more stable areas. The
importance of such social institutions as the family and the school in de-
veloping internal social control among children and youth (i.e., for the
socialization process) has often been stressed, whereas less attention has
been given to the interplay between the family and the school in this task in
different types of residential areas. It is most likely that the family and the
school work under very different conditions in different types of residential
areas and that the interplay between the two may show differences depend-
ing on the general characteristics of the area and its population. This
appears to be an important area in need of further research.
(in a broad sense) to socially deviant persons (i.e., high participation in risk
milieus such as public entertainment), or to have a job that means coming
into frequent contact with socially deviant persons (see Chaps. 4, 5, and 8).
It is noteworthy in this context that the residential pattern for victims of
violence shows a rather close relationship to that of the offenders (see
Chap. 8).
In a general way, one may state that the risk of victimization for expres-
sive crimes of violence and vandalism (and some types of instrumental
violence) is dependent on the degree to which the individual has risk
relationships and the frequency with which he or she is confronted with
risk meetings. Risk relationships may loosely be defined as relationships
to socially deviant persons, and risk meetings as meetings in socially un-
stable public milieus (e.g., public entertainment areas). The frequency of
risk relationships and risk meetings may be assumed to be a consequence
of life-style.
The risk of being subjected to noninteractional instrumental types of
crime (predominantly crimes of theft) appears to be both more evenly dis-
tributed in the population than the victimization risk for expressive forms
of violence and vandalism and highly dependent on the area of residence.
In the case of noninteractional instrumental crimes, it is perhaps better to
relate the victimization risk to living in risk areas rather than to having risk
relationships and risk meetings. However, the evidence suggests that differ-
ent types of areas may show different risks for being subjected to different
types of noninteractional instrumental crimes (see Chaps. 4, 5, and 8).
tions for different types of external social control and the amount of risk
relationships, risk meetings, and opportunities for noninteractional in-
strumental types of crimes are likely to show significant variablity among
times and among areas of a city (see chap. 8). This is probably the key to
understanding city area variations in the rate and structure of crime and
victimization.
References
246
References 247
Reiss & M. Tonry (Eds.), Communities and Crime. Crime and Justice 8.
Chicago: The University of Chicago Press.
Bursik, R.J.
1988 Social disorganization and theories of crime and delinquency: Problems
and prospects. Criminology, 26.
Burt, C.
1944 The young delinquent. London: University of London Press.
Capone, D.L., & Nichols, IL, W.W.
1976 Urban structure and criminal mobility. American Behavioral Scientist, 20.
Carlssson, G.
1972 Samhiillsmiljo och rumslig fordelning. IN SOU, 1972:76.
Carlsson, G.
1975 Kriminalitetsniva och belastningsfordelningar. BrottsfOrebyggande radet.
PM 1975:1.
Carter, R.L., & Hill, K.Q.
1979 The criminal's image of the city. New York: Pergamon Press.
Castle, LM., & Gittus, E.
i961 The distribution of social defects in Liverpool. In G.A. Theodorson
(Ed.), Studies in human ecology (pp. 415-429). New York: Row, Peter-
son and Company.
Centralforbundet for socialt arbete
1941 Ligabrottsligheten. Kooperativa forbundets for1ag.
Chilton, R.J.
1964 Continuity in delinquency area research: A comparison of studies from
Baltimore, Detroit and Indianapolis. American Sociological Review, 29.
Chilton, R.J., & Dussich, J.P.J.
1974 Methodological issues in delinquency research: Some alternative analyses
of geographically distributed data. Social Forces, 53.
Christie, N.
1975 Hvor teU et samfunn? Oslo: Universitetsforlaget.
Christie, N., Andenaes, J., & Skirbeckk, S.
1965 A study of self-reported crime. Scandinavian Studies in Criminology, 1.
Clarke, R.V.G., & Mayhem, P.
1980 Designing out crime. London: Home Office Research Unit Publications.
Clinard, M.B.
1974 Sociology of deviant behavior. New York: Holt, Reinhart & Winston.
Clinard, M.B.
1978 Cities with little crime. London: Cambridge University Press.
Cohen, L.E., & Felson, M.
1979 Social change and crime rate trends: A routine activity approach. Amer-
ican Sociological Review, 44.
Cohen, S.
1984 Sociological approaches to vandalism. In C. Levy-Leboyer (Ed.), Vandal-
ism: Behaviour and motivations (pp. 51-61). Amsterdam: Elsevier Sci-
ence Publications.
Conklin, J.E.
1972 Robbery and the criminal justice system. Philadelphia: J.B. Lippincott
Company.
Crutchfield, R.D., Geerken, M.R., & Gove, W.R.
1982 Crime rate and social integration. Criminology, 20.
References 249
Curtis, L. A.
1974 Criminal violence. Lexington, MA: Lexington Books.
Davidson, R.N.
1981 Crime and environment. London: Croom Helm.
Deardorff, C.M., Melges, F.T., Hout, C.N. & Savage, D.J.
1975 Situations related to drinking alcohol. Journal of Studies on Alcohol, 36.
van Dijk, J.J.M., & Vivanen, A.C.
1978 Criminal Victimization in the netherlands. The Hauge: Ministry of Justice.
Dobash, R., & Dobash, R.
1979 Violence against wives. London: Open Books.
Dobash, R., & Dobash, R.
1984 The nature and antecedents of violent events. British Journal of Criminol-
ogy,24.
Downes, D.
1958 The delinquent solution. London: Routledge & Kegan Paul.
Ds S 1980:11
1980 rungt Narkotikamissbruk. Socialdepartementet.
Duncan, O.D., & Duncan, B.
1955a A methodological analysis of segregation indexes. American Sociological
Review, 20.
Duncan, O.D., & Duncan, B.
1955b Residential distribution and occupational stratification. The American
Journal of Sociology, 60.
Duner, A., & Haglund, B.
1974 Tonarspojkar och bratt. Stockholm: Utbildningsf6rlaget.
Dunn, C.S.
1976 The Patterns and Distribution of Assault Incident Characteristics Among
Social Areas (Analytic Report No. 14). Washington, D.C.: U.S. Depart-
ment of Justice.
Edfeldt, A.W.
1985 Allt vad I goren . ... Stockholm: Proprius fOrlag.
Elmhorn, K.
1969 Faktisk brottslighet bland skolbarn. SOU 1969: l.
Farley, J.E., & Hansel, M.
1981 The ecological context of urban crime. Urban Affairs Quarterly, 17.
Farrington, D.P.
1983 Offending from 10 to 25 years of age. In K.T. Van Dusen & S.A. Mednick
(Eds.), Prospective studies on crime and delinquency (pp. 17-37). Boston:
Kluwer-Nijhoff.
Farrington, D.P.
1984 Measuring the natural history of delinquency and crime. In Advances in
the behavioral measurement of children: Voll (pp. 217-263).
Farrington, D.P.
1986a Age and crime. In Tonry & Morris (Eds.), Crime and Justice 7. Chicago:
The University of Chicago Press.
Farrington, D.P.
1986b What have we learned from major longitudinal studies? In D.P. Farring-
ton, & L.E. Ohlin & J.Q. Wilson (Eds.), Understanding and controlling
crime (pp. 25-64). New York: Springer-Verlag.
250 References
Farrington, D.P.
1988a Advancing knowledge about delinquency and crime: The need for a coor-
dinated program of longitudinal research. Behavioral Science & the Law,
6.
Farrington, D.P.
1988b Social, psychological and biological influences on juvenile delinquency
and adult crime. In W. Buikhuisen & S.A. Mednick (Eds.), Explaining
Criminal Behaviour (pp. 68-89). Leiden: E.J. Brill.
Farrington, D.P., & Wikstrom, P.-O.
1990 Age, crime and criminal careers in London and Stockholm. (Project
Metropolitan Research Reports). Stockholm: University of Stockholm,
Department of Sociology (in press).
Fattah,E.A.
1989 Victims and victimology: The facts and the rhetoric. International Review
of Victimology, I.
Feeney, F.
1986 Robbers as decision-makers, pp. 53-72 in Cornish D.B. & Clarke R.V.
(Eds), The Reasoning Criminal. New York: Springer-Verlag.
Felson, M.
1986 Linking criminal choice, routine activities, informal control, and criminal
outcomes. In D.B. Cornish & R.V. Clarke (Eds.), The Reasoning Crimi-
nal (pp. 119-128). New York: Springer-Verlag.
Felson, M.
1987 Routine activities and crime in the developing metropolis. Criminology,
25.
Felson, R.B., & Steadman, H.J.
1983 Situational factors in disputes leading to criminal violence. Criminology,
21.
Fox, J.A., & Tracy, P.E.
1988 A measure of skewness in offense distributions. Journal of Quantitative
Criminology, 4.
Garofalo, J.
1987 Reassessing the life-style model of criminal victimization. In M.R. Gott-
fredson & T. Hirschi (Eds.), Positive criminology (pp. 23-42). Beverly
Hills, CA: Sage Publications.
Gibbs, J.P., Erickson, M.L.
1976 Crime rates of American cities in an ecological context. American Journal
of Sociology, 82.
Glaser, D.
1974 The classification of offenses and offenders. In D. Glaser (Ed.), Hand-
book of criminology (pp. 45-83). Chicago: Rand McNally.
Gordon, R.A.
1967 Issues in the ecological study of delinquency. American Sociological Re-
view, 32.
Gorsuch, R.L.
1974 Factor analysis. Philadelphia: W.B. Saunders.
Gottfredson, M.R.
1984 Victims of Crime: The Dimension of Risk. Home Office Research Study
No. 81. London: Home Office.
References 251
in urban communities. In G.D. Suttles & M.N. Zald (Eds.), The challenge
of social control (pp. 230-242). Norwood, NJ: Ablex.
Janson, C.-G.
1953 Slutrapport for kriminologiska institutets undersokning av brottslighetens
regionala fordelning. Unpublished manuscript.
Janson, c.-G.
1970 Juvenile delinquency in Sweden. Youth and Society, 2.
Janson, C.-G.
1971 A preliminary report on Swedish urban spatial structure. Economic
Geography,2(Suppl.).
Janson, c.-G.
1977 The Handling of Juvenile Delinquency Cases (project Metropolitan Re-
search Reports No. 17). Stockholm: University of Stockholm, Depart-
ment of Sociology.
Janson, C.-G.
1980 Factorial social ecology: An attempt at summary and evaluation. Ann.
Rev. Sociol, 9.
Janson, C.-G.
1982 Delinquency among metropolitan boys (Project Metropolitan Research
Reports No. 17). Stockholm: University of Stockholm, Department of
Sociology.
Janson, C-G.
1984 Project Metropolitan: A Presentation and Progress Report (Project Metro-
politan Research Reports No. 21.) Stockholm: University of Stockholm,
Department of Sociology.
Jeffery, C.R.
1977 Crime prevention through environmental design. Beverly Hills, CA: Sage
Publications.
Johansson, I.
1987 Stor-Stockholms bebyggelsehistoria. Gidlunds forlag.
Johnston, R.J.
1984 City and society: An outline for urban geographgy. London: Unwin
Hyman.
Jonassen, C.T.
1949 A re-evaluation and critique of the logic and some methods of Shaw and
McKay. American Sociological Review, 14.
Jones, T., Maclean B., & Young, J.
1988 The Islington crime survey. Aldershot, England: Gower.
Jonsson, G.
1969 Det Sociala Arvet. Stockholm: Tidens forlag.
Knutsson, J., & Kuhlhom, E.
1979 Missbruk och brott-en litteraturoversikt. Unpublished manuscript. Stock-
holms Universitet, Sociologiska Inst., Stockholm.
Knutsson, J.
1980 Om Bostadsinbrott-polisanmalda bostadsinbrott i Stockholm lir 1977.
Svensk juristtidning.
Knutsson, J.
1984 Polisen och brottspreventionen. Stockholms Universitet, Kriminologiska
Inst., Stockholm.
References 253
Kornhauser, R.R.
1978 Social sources of delinquency. Chicago: The University of Chicago
Press.
Komer, S., Ek, L., & Berg, S.
1984 Deskriptiv Statistik. Lund: Studentlitteratur.
Kuhlhorn, E.
1984 Den svenska vdldsbrottsligheten. (BRA. Rapport 1984:1). Stockholm:
Liber forlag.
Kuhlhorn, E.
1990 Victims and offenders of criminal violence. Journal of Quantitative Crimi-
nology,6.
Lagerberg, D.
1982 Du skall icke sid. Om fysisk ock psykisk misshandel av barn. Stockholm:
Liber forlag.
Lander, B.
1954 Towards an understanding of juvenile delinquency. New York: Columbia
. University Press.
LeBlanc, M., & Frechette, M.
1989 Male criminal activity, from childhood through youth; multilevel and de-
velopment perspectives. New York: Springer-Verlag.
Lenke, L.
1974 Vdldsbrottsligheten i Stockholm. Unpublished manuscript, Stockholms
Universitet, Kriminologiska Inst., Stockholm.
Lenke, L.
1978 Risken for vdldsbrott i Sverige. Unpublished manuscript, Stockholms Uni-
versitet, Kriminologiska Inst., Stockholm.
Loeber, R., & LeBlanc, M.
1990 Toward a developmental criminology. In M. Tonry & N. Morris (Eds.),
Crime and Justice, 12. Chicago.
Lofland, L.H.
1973 A world of strangers. New York: Basic Books.
Lundsgaarde, H.P.
1977 Murder in space city. New York: Oxford University Press.
Mawby, R.
1979 Policing the city. Westmead, England: Saxon House.
Maxfield, M.G.
1987 Lifestyle and routine activity theories of crime: Empirical studies of vic-
timization, delinquency, and offender decision-making. Journal of
Quantitative Criminology, 3.
Mays, J.B.
1963 Crime and the Social Structure. Faber and Faber Ltd. London.
McCaghy, C.H., Giordano, P.C., & Henson, T.K.
1977 Auto theft: Offender and offense characteristics. Criminology, 15.
McClintock, F.H.
1963 Crimes of Violence. London: Macmillan.
McClintock, F.H.
1974 Phenomenological and Contextual Analysis of Criminal Violence. Col-
lected studies in criminological research: Vol. Xl. Violence in Society (pp.
127-176). Strasbourg: Council of Europe.
254 References
McClintock, F.H.
1981 Criminological Aspects. In Cities and Criminality (pp. 6-21). Report from
the Tenth International Congress on Social Defense.
McClintock, F.H., & Gibson, E.
1961 Robbery in London. London: Macmillan.
McClintock, F.H., & Wikstrom, P.-O.
1987 Valdsbrott i Sverige och Skottland. In Brottsutvecklingen 1987 (BRA
Forskning 1987:5). Stockholm: Liber forlag.
McIver, J.P.
1981 Criminal mobility. In S. Hakim & G. Regnert (Eds.), Crime Spillover.
Beverly Hills, CA: Sage Publications.
Megargee, E.I.
1982 Classifying criminal offenders. Beverly Hills, CA: Sage Publications.
Morris, T.
1958 The criminal area. London: Routledge & Kegan Pau!'
Mowrer, E.R.
1938 The isometric map as a technique of social research. The American Jour-
nal of Sociology, 44.
Nettler, G.
1974 Explaining Crime. New York: McGraw-Hill.
Newman, O.
1972 Defensible space. London: Architectural Press.
Nilsson, L.V.
1988 Antastande, fridstdrande och terroriserande. In Brottsutvecklingen.
(BRA forskningen 1988:2).
Olofsson, B.
1971 Vad var det vi sa! Om kriminellt och konformt beteende bland skolpojkar.
Stockholm: UtbildningsfOrlaget.
Olson, P.
1982 Urban neighborhood research: Its development and current focus. Urban
Affairs Quarterly, 17.
Partanen, P.
1981 Skadlig behandling av barn i Botkyrka kommun. Unpublished manu-
script, Stockholms universitet, Sociologisk Inst., Stockholm.
Persson, L.G.W.
1976 Inbrottstjuvar i Stockholm. Svensk Juristtidning.
Persson, L.G.W.
1977 Offer far tillgrepp, skadegarelser och viild-en redovisning av 1974 iirs
offerundersokning (SCB Promemorior 1977:7).
Persson, L.G.W.
1980 Hidden criminality. Stockholm: University of Stockholm, Department of
Sociology.
Persson, L.G.W.
1981 Viildtiikt (SOU 1981:64). Stockholm: Liber fOriag.
Peterson, R.A., Pittman, D.J., & O'Neal, P.
1962 Stabilities of deviance: A study of assaultive and non-assaultive offenders.
Journal of Criminal Law, Criminology and Police Science, 53.
Phillips, P.D.
1980 Characteristics and typology of the jounrney to crime. In D.E. Georges-
References 255
Abeyie & K.D. Harries (Eds.), Crime: A spatial perspective (pp. 167-
180). New York: Columbia University Press.
Pittman, D.J., & Handy, W.
1964 Patterns in criminal aggravated assault. The Journal of Criminal Law,
Criminology and Polic Science, 55.
Polk, K.
1967 Urban social areas and delinquency. Social Problems, 14.
Popenoe, D.
1977 The suburban environment. Chicago: The University of Chicago Press.
Porkny, A.D.
1965 Human violence: A comparison of homicide, aggravated assault, suicide
and attempted suicide. The Journal of Criminal Law, Criminology and
Police Science, 56.
Powers, R.J., & Kutash, I.L.
1978 Substance-induced aggression. In I.L. Kutash, & S.B. Kutash & L.B.
Schlesinger (eds.), Perspectives on Murder and aggression (pp. 317-342).
San Francisco: Jossey-Bass.
Pratt, M.
1980 Mugging as a social problem. London: Routledge and Kegan Paul.
Prus, R.C.
1978 From barroms to bedrooms. In M.A.B. Gammon (Ed.), Violence in
Canada (pp. 51-73). Toronto: Methuen.
Pyle, G.F.
1974 The spatial dynamics of crime (Research Paper No. 159). Chicago: The
University of Chicago, Department of Geography.
Reiss, A.J.
1971 The Police and the public. New Haven, CT: Yale University Press.
Reiss, A.J.
1986 Why are communities important in understanding crime? In A.J. Reiss &
M. Tonry (Eds.), Crime and justice 8: Communities and crime. Chicago:
University of Chicago Press.
Reiss, A.J., & Rhodes, L.E.
1961 The Distribution of Juvenile Delinquency in the Social Class Structure.
American Sociological Review, 26.
Reiss, A.J., & Tonry, M.
1986 Crime and justice 8: Communities and crime. Chicago: The University of
Chicago Press.
Rhodes, W.M., & Conly, C.c.
1981 Crime and mobility: An empirical study. In P.J. Brantingham & P.L.
Brantingham (Eds.), Environmental criminology (pp. 167-188). Beverly
Hills, CA: Sage Publications.
Rojek, D.G., & Erickson, M.L.
1982 Delinquent careers: A further test of the escalation model. Criminology,
20.
Roos, H.-E.
1986 Vandalism i storstad och glesbygd. Stockholm: Byggforsknings-
radetlLiber.
Rosenblatt, E., & Greenland, C.
1979 Female crimes of violence. Canadian Journal of Criminology and Correc-
tions,16.
256 References
Sampson, R.J.
1986a Neighborhood family structure and the risk of personal victimization. In
J.M. Byrne & R.J. Sampson (Eds.), The social ecology of crime. New
York: Springer-Verlag.
Sampson, R.J.
1986b Crime in cities: The effects of formal and informal social control. In A.J.
Reiss & M. Tonry (Eds.), Crime and justice 8: Communities and crime
(pp. 271-331). Chicago: University of Chicago Press.
Sarnecki, J.
1983 Fritid och brottslighet. (BRA forskning 1983:7.) Stockholm: Liber fi:irlag.
SCB
1981 Offer for vl1lds- och egendomsbrott. (Rapport Nr. 24.) Statistiska Central-
byrlm. Levnadsforhallanden 1978.
Schmid, c.P.
1960a Urban crime areas, Part I. American Sociological Review, 25.
Schmid, c.P.
1960b Urban crime areas, Part II. American Sociological Reivew, 25.
Schuerman, L., & Kobrin, S.
1986 Community Careers in Crime. In A.J. Reiss & M. Tonry (Eds.), Crime
and justice 8: Communities and crime (pp. 67-100). Chicago: The Uni-
versity of Chicago Press.
Schuessler, K.
1962 Components of variation in city crime rates. Social Problems, 9.
Schuessler, K., & Slatin G.
1964 Sources of variation in U.S. city crime, 1950 and 1960. Journal of Re-
search on Crime and Delinquency, 1.
Schwirian, K.P.
1974 Some recent trends and methodological problems in urban ecological re-
search. In K.P. Schwirian (Ed.), Comparative urban structure (pp. 3-31).
Lexington, MA: D.C. Heath and Company.
Schwirian, K.P.
1983 Models of neighborhood change. Ann. Rev. Social, 9.
Shaw, C.R.
1931 The natural history of a delinquent career. Chicago: The University of
Chicago Press.
Shaw, C.R., & McKay, H.
1949 Rejoinder. American Sociological Review, 14.
Shaw, C.R., & McKay, H.
1969 Juvenile delinquency in urban areas. Chicago: The University of Chicago
Press.
Shelley, L.
1980 The geography of Soviet criminality. American Sociological Review, 45.
Shoman, S.G., Ben-David, S., & Rahav, G.
1975 Interactions in violence. Human Relations, 27.
Skogan, W.G.
1978 The changing distribution of big-city crime. Urban Affairs Quarterly, 13.
Smith, D.A.
1986 The Neighborhood Context of Police Behavior. In A.J. Reiss & M. Tonry
References 257
(Eds.), Crime and justice 8: Communities and crime (pp. 313-341). Chi-
cago: The University of Chicago Press.
Socialystyrelsen
1969 En utredning om barnmisshandel. Socialstyrelsen redovisar Nr. 9.
Sparks, R.F., Genn, H.G., & Dodd, D.J.
1977 Surveying victims. London: John Wiley & Sons.
Stattin, H., Magnusson, D., & Reichel, H.
1986 Criminality from childhood to adulthood: A longitudinal study of criminal
behaviour. Part 1: Criminal activity at different ages. (Report No. 63).
University of Stockholm.
Steinmetz, C.H.D.
1981 Victimological risk analysis. The Hague: Ministry of Justice.
Straus, M.A., Gelles, R., & Steinmetz, S.
1980 Behind closed doors. New York: Anchor Books.
Suikkila, J.
1983 Nagra synpunkter pa alkoholbruk och brottslighet bland ungdomar i Sver-
ige och Finland, In U.-B. Eriksson & H. Tham (Eds.), Utliinningarna och
brottslighteten (pp. 190-214). BRA Forskning 1983:4. Stockholm: Liber
fOrlag.
Sveri, K.
1960 Kriminalitet og alder. Stockholm: Almqvist & Wiksell.
Sveri, K.
1961 Criminality and age. Acta Sociologica, 4.
Sveri, K.
1965 Group activity. Scandinavian Studies in Criminology, 1.
Sveri, K.
1974 Brottslighetens volym och struktur. In A. Daun, B. Borjeson, & S. Ahs
(Eds.), Samhiillsforiindringar och brottslighet. Stockholm: Tidenl
Folksam.
Taub, R.P., Taylor, D.G., & Dunham, J.D.
1984 Paths of neighborhood Change: Race and crime in urban America. Chica-
go: The University of Chicago Press.
Taylor, S.P., & Leonard, K.E.
1983 Alcohol and human physical aggression. In G.G. Russel & E.!. Donner-
stein (Eds.), Theoretical and empirical reviews (pp. 77-101). New York:
Academic Press.
Tham, H.
1978 Brottslighet och levnadsniva. Stockholm: Liber forlag.
Theil, H.
1972 Statistical decomposition analysis. Amsterdam: North-Holland.
Theodorson, G.A.
1961 Studies in human ecology. Evanston, IL: Row, Peterson and Company.
Thornberry, T.P., & Farnworth, M.
1982 Social correlates of criminal involvement: Further evidence on the rela-
tionship between social status and criminal behavior. American Socio-
logical Review, 47.
Timms, D.W.G.
1971 The urban mosaic. Cambridge: Cambridge University Press.
258 References
Wikstrom, P.-O.
1983 Social klass och brottslighet. (BRA appropa No.3).
Wikstrom, P.-O.
1985 Everyday violence in contemporary Sweden: Ecological and situational
aspects. (BRA Report No. 15). Stockholm: Liber forlag.
Wikstrom, P. -0.
1987a Patterns of crime in a birth cohort: Sex, age and social class differences.
(Project Metropolitan Research Reports No. 24). Stockholm: University
of Stockholm, Department of Sociology.
Wikstrom, P.-O.
1987b Brott och stadsmilko. (BRA Forskning 1987:3). Stockholm: Liber forlag.
Wikstrom, P.-O.
1987c Villdsbrott och alkohol: Forskning & fakta. Stockholm: CAN.
Wikstrom, P.-O.
1989a Housing and offending: The individual-level relationship in childhood and
youth. In Project Metropolitan Research Report No. 27 (pp. 42-48).
Stockholm: University of Stockholm, Department of Sociology.
Wikstrom, P.-O.
1989b Homicide in Stockholm, 1970-1987. (Unpublished manuscript). Stock-
holm: University of Stockholm, Department of Criminology.
Wikstrom, P.-O.
1990 Age and crime in a stockholm cohort. Journal of Quantitative Criminolo-
gy,6.
Wilkins, L.T.
1964 Social deviance. London: Tavistock.
Winchester, S. & Jackson, H.
1982 Residential burglary. (Home Office Research Study No. 74). London:
Home Office.
Wirth, L.
1938 Urbanism as a way of life. American Journal of Sociology.
Wolfgang, M.E.
1958 Patterns in criminal homicide. Philadelphia: University of Philadelphia
Press.
Wolfgang, M.E., Figilio, R.M., & Sellin, T.
1972 Delinquency in a birth cohort. Chicago: University of Chicago Press.
Wolfgang, M.E., Thornberry, T.P., & Figlio, R.M.
1987 From boy to man, from delinquency to crime. Chicago: The University of
Chicago Press.
Akerstrom, M.
1983 Crooks and squares. Lund: Studentlitteratur.
Appendix A
The Stockholm 1982 Crime Survey
What is called the Stockholm 1982 Crime Survey in this book consists in
fact of several different studies on urban crime, directed by the author and
conducted at the Department of Criminology, University of Stockholm,
and at the Research Unit of the National Council for Crime Prevention,
Sweden. Although these are separate research projects, they have all been
designed as to be compatible with the first original study of crime in Stock-
holm, 1982, which included homicides, assaults, robberies, vandalism, and
residential burglaries (see Wikstrom, 1987b). The latter research projects
have added violence to a public official, rapes, molestations, l nonresiden-
tial burglaries, and thefts of and from cars to those crime types originally
studied. All in all, the Stockholm 1982 Crime Survey includes a detailed
study of the investigation records of samples of police reported crimes of
violence (including molestation), vandalism, and serious property crimes
(burglaries and thefts of and from cars). In total the sample covers 7,575
cases (see Table A.l).
The Samples
The sampling of crimes was basically systematic (e.g., every third, every
fifth etc.) crime, depending on category of crime in question, was included.
The exceptions are serious assaults and rapes, for which all cases during the
sampling period (see below) were included. A systematic sample can be
viewed as a random sample if the list from which the sample is drawn is
ordered so that the values on the different included variables appear ran-
domly. The list from which the samples were drawn were the police records
ordering number list, in which the crimes are ordered succsesively, from 1
to the maximum number of crimes reported during the year in question, by
the point in time they are reported to the police. There is no reason to
1. The study on molestation was made by L. Nilsson (1988) as exam work for a law
degree.
260
The Samples 261
TABLE A.I. Number of studied crimes, number of total crimes (sample weighted to
I-year figures) and percentage of all reported crimes included in study (The Stock-
holm 1982 Crime Survey).
No. of studied Total crimes a % of all crimes
Crime crime (I-year figure) reported studied
suspect that the values of the variables included (e.g., sex, age, place of
crime) will show any systematic variation with the ordering number (other
than, of course, month of crime and similar variables).
However, in addition to the systematic samples, all remaining cases with
a known suspect were included for those crimes with low clearance rates.
The obvious reason for that was to get a sufficient number of cases with
offenders to analyze. These "extra samples" were made for robberies, res-
idential burglaries, nonresidential burglaries, and thefts of and from cars.
Finally, the period from which the sample was drawn differed for the
types of included crimes. For serious and petty assault, robberies, vandal-
ism, and residential burglaries the sample was taken from the cases re-
ported January-July 1982. For violence toward a public official, molesta-
tion, nonresidential burglary, thefts of and from cars, the sample was taken
from all cases reported during 1982 (with the exception of molestation, for
which the sample was drawn from all cases reported during 1983). Finally,
for rapes, all cases reported during 2 years were included (i.e., 1982 and
1983). There is no specific reason for the differences in sampling periods
other than that different sampling periods were employed in the different
research projects making up the Stockholm 1982 Crime Survey. It was, as
far as possible, investigated whether the restriction of the sampling period
to parts of the year for some types of the crime would significantly affect
the results in comparison with whether the sample had been drawn from a
full year. These tests showed, however, that the results are not significantly
262 Appendix A
affected by the restriction in the sampling period, and that is so because the
major seasonal differences in crime circumstances are between the warm
and cold part of the year, and a part of both the warm and the cold period
of the year was included (see Wikstrom, 1987b, pp. 45-53).
The fact that cases with a known offender have been oversampled for
some of the included types of crimes means, of course, that for these
crimes we have more known offenders than what applies to all reported
cases. Moreover, insofar as cases with a known offenders deviates, as re-
gards the crime circumstances and victim characteristics, from those with
unknown offenders, a bias is introduced in the description of crime cir-
cumstances and victim characteristics. However, the extent of the possible
bias introduced has to be viewed in reference to the size of the extra sam-
ples, which are rather small compared to the systematic samples.
In this appendix, the legal definitions, as stated in the Swedish Penal Code,
are given for the crimes included in the Stockholm 1982 Crime Survey. The
source for this section is the 1986:2 report by the National Council for
Crime Prevention, Sweden, which is an English translation of the Swedish
penal code.
264
Robbery 265
an official act, or for the purpose of taking revenge for such act, he shall be
sentenced for violence or threat to public servant to imprisonment for at
most 4 years or, if the crime is petty, to pay a fine or to imprisonment for at
most 6 months. The same shall apply if a person attacks someone who has
previously exercised authority for something he had done or failed to do
while in office. 1
Rape
If a person, by violence or by threat involving or appearing to the
threatened person as imminent danger, forces the latter to copulate or
have other comparable sexual intercourse, he or she shall be sentenced for
rape to imprisonment for at least 2 and at most 6 years. Rendering the
person unconscious or putting the person in other such state shall be re-
garded as equivalent to violence.
In view of the nature of the violence or the threat and the circumstances in other
respects the crime is considered less grave, a sentence to imprisonment for at most
4 years shall be imposed.
If the crime is grave, a sentence to imprisonment for at least 4 years and at
most 10 years shall be imposed for aggravated rape. In judging whether the
crime was grave, special attention shall be paid to whether the violence
involved a danger to life or whether the offender inflicted serious injury or
serious illness or otherwise displayed particular brutality.
Robbery
A person who steals by means of violence against another or by means of
threat implying or appearing to the threatened person as imminent danger
or who, after committing a theft being caught in the act, resists by such
violence or threat the one who wants to recover the stolen property, shall
be sentenced for robbery to imprisonment for at least 1 and at the most 6
years. The same shall apply if a person by such violence or threat forces
another to do or not do something resulting in gain to the offender and loss
to the person so forced or to someone he represents. Rendering a person
unconscious or similarly helpless is eqvivalent to violence.
If the crime is of a less serious nature in respect of the violence, threat or other
circumstances, the sentence shall not be for robbery but for such other offense as
the procedure entails.
1. Only the cases toward a public official involving actual violence have been in-
cluded in this study.
266 Appendix B
Molestation
A person who manually molests or by discharge of a firearm, throwing of stones,
making loud noise, or other heedless conduct harrasses another, shall be sentenced
for molestation to pay a fine or to imprisonment for at most 6 months.
Vandalism
A person who destroys or damages real property or chattels to the detri-
ment of another's right thereto, shall be sentenced for inflicting damage to
pay a fine or to imprisonment for at most 6 months.
If, considering the minimal harm and other circumstances of the act, the
crime is regarded as petty, a fine shall be imposed for trespassing.
If the crime is regarded as grave, imprisonment for at most 4 years shall be imposed
for inflicting gross damage. In judging the gravity of the crime, special attention
shall be paid to whether the act had given rise to substantial danger to anyone's life
or health or damage had been done to something of great cultural or economic
importance or else keenly felt.
* Burglary, and thefts from cars, do not appear as special types of crimes in the
Swedish Penal Code, but under the legal code for crimes of theft. Residential burg-
laries are normally considered to be grand thefts.
Crimes of Theft 267
If the crime is regarded as grave, imprisonment for at least 6 months and at most 6
years shall be imposed for grand theft. In judging the gravity of the crime, special
attention shall be paid to whether it was aimed at an object which someone carried
on his person, whether the offender was equipped with a weapon, explosive, or
other similar means or whether the act otherwise was of a specially dangerous or
unscrupulous nature, directed at substantial values or involved a keenly felt loss.
If a person unlawfully takes or uses a motor vehicle or other motor-driven
conveyance belonging to another, he shall, unless the crime is punishable
under earlier provisions of this chapter, be sentenced for vehicle theft to
imprisonment for at most 2 years or, if the crime is of a petty nature, to pay
a fine.
If the crime is grave, imprisonment for at least 6 months and at most four
years shall be imposed.
AppendixC
Definitions of Variables Used in the
Factor Analysis of the Stockholm
Urban Structure
In this appendix the definitions are given of the variables used in the factor
analysis of Stockholm urban structure presented in Chapter 6.
Variable Comment
Ward Characteristics
Distance from city center Distance in hundreds of meters from the Sergei square to
the midpoint of the built-up part of the ward
Workplaces in commerce Number of persons working in an area in shops, service
institutions, hotels, restaurants, etc.
Manufacturing workplaces Number of persons working in an area in manufacturing
Workplaces Persons working in an area divided by the sum of persons
working and residing in the area (this measure will be
100 if there are no residences in an area and 0 if there are
no workplaces)
Demographic characteristics
Young persons Percentage of the residential population aged 0-14
Old persons Percentage of the residential population aged 60 or older
Females Percentage of females 15 years and older of the residential
population aged 15 years and older
Foreigners Percentage of the residential population with foreign
citizenship
Household characteristics
Household size Mean
Overcrowding Percentage of all households with more than two persons
per room, kitchen and one room excluded
Households with children Percentage of households with members aged 0-17 years.
Single working parents Households with single working parents with children aged
0-12 years
Housing characteristics
Houses Percentage of all residences in area in one- and two-family
houses
268
Appendix C 269
Socioeconomic characteristics
Incoqte Mean
Income heterogeneity Mean-median
Blue-collar workers Percentage of residential population employed in manufac-
turing
Females gainfully employed Percentage of females with an income of 20,000 Swedish
Crowns or more of all females 15 years and older
Public welfare assistance Percentage of residential population receiving public wel-
fare assistance
Research in Criminology
continued