Comparing predictors of victimisation and attitudes towards crime across five European countries

advertisement
Comparing predictors of victimisation
and attitudes towards crime across five
European countries
Peter Baudains
UCL Department of Security and Crime Science
Outline
• Introduction: The MARGIN project and motivation for
this study.
• Theory: The role of victimisation, vulnerability, and
neighbourhoods in perceived insecurity.
• Data: Victimisation surveys in five European countries.
• Results: Identification of consistent and inconsistent
cross-national predictors.
• Conclusions
MARGIN: Tackle insecurity in marginalized areas
http://marginproject.eu
• Horizon 2020 project running from May 2015 - April 2017.
• Objective:
• Understand the emergence of perceived insecurity within different demographic groups across
Europe.
• Policies for the reduction of insecurity within marginalised communities.
• Methods:
• Quantitative analysis of existing data sources;
• Qualitative analysis within 10 example neighbourhoods in five cities across Europe;
• Development and implementation of survey to capture the emergence of insecurity in
marginalised groups.
• With UCL colleagues Spencer Chainey, Aiden Sidebottom, Kate Bowers and Richard Wortley. In
addition to seven partnering institutions across Europe.
The utility of cross-national comparisons
• Country-level effects on victimisation and
perceived insecurity.
• Which countries do a better job of reducing
p e r c e i v e d i n s e c u r i t y, p a r t i c u l a r l y i n
marginalised communities?
• Country differences might be due to:
• Implementation of good policies (can they
be replicated?)
• Variation in the performance of police
forces or other institutions
• Cultural/social/political differences
• Victimisation and fear of crime is understudied
in many countries. Do we reach the same
theoretical conclusions in these contexts?
Image: http://www.d-maps.com/carte.php?num_car=30083&lang=en
The International Crime Victimisation Survey (ICVS)
• The ICVS has been conducted every 4
to 5 years since 1989 in over 80
countries and forms the basis of many
cross-national comparisons relating to
victimisation and attitudes towards
crime.
• Designed specifically to enable crossnational comparisons.
• Limited in sample size within each
participating country.
• Also limited in the scope of questions
asked.
• In spite of best intentions, some (minor)
cross-national differences still exist.
Van Dijk, J. (2008). The World of Crime. Sage Publications, Thousand
Oaks, CA.
National victimisation surveys
• On the other hand, national victimisation surveys are:
• Routinely employed (often yearly)
• Have much larger sample sizes
• Have a much broader range of questions relating
to different aspects of perceived insecurity and
victimisation (often > 100 questions).
National victimisation surveys
“How safe do you feel in your
neighbourhood/municipality?”
• On the other hand, national victimisation surveys are:
• Routinely employed (often yearly)
• Have much larger sample sizes
• Have a much broader range of questions relating
to different aspects of perceived insecurity and
victimisation (often > 100 questions).
Catalonia (unweighted)
Responses
Percentages
(neighborhood)
Percentages
(municipality)
10 (very safe)
3.0%
3.3%
5.6%
5.3%
8
19.9%
19.2%
7
23.0%
22.1%
6
16.7%
17.8%
5
15.5%
16.9%
4
6.4%
7.0%
3
4.1%
3.8%
2
2.3%
1.9%
1
0.8%
0.7%
0 (very unsafe) 2.9%
1.9%
• But are not designed from the outset for cross-national
9
comparison:
“Do you personally feel unsafe in
your neighbourhood or village?”
France
(unweighted)
Responses
Percentages
Often
3.3%
Sometimes
9.6%
Hardly ever
9.9%
Never
77.1%
Research strategy
• In order to account for small differences in question wording, we find
predictors of insecurity and victimisation within countries using multivariate
models.
• We then compare predictors across countries.
• Research question becomes:
Does the variation in how different types of people respond vary
across countries?
• And, additionally, we need to consider:
Can some of this variation be attributed to country-level differences
(as opposed to differences in survey design)?
• Differences in survey design and question wording will not affect
conclusions of multivariate models if these differences are not correlated
with the predictors.
Theory: indicators of insecurity and victimisation
• (Repeat-) victimisation hypothesis:
Individuals who have been the victim of crime are more likely to:
i) experience further victimisation, and
ii) have increased feelings of perceived insecurity.
• Vulnerability hypothesis:
Individuals who:
1) have a higher exposure to risk;
2) are likely to incur more serious consequences should they be the victim of a
crime; and
3) are less likely to be able to control a situation in which they become a victim;
are more likely to develop higher levels of fear of crime.
• Environmental effects:
• Exposure to risk can arise from the environment in which the respondent finds
themselves. Neighbourhoods with more disorder and less social organisation
are likely to exhibit higher levels of victimisation and fear of crime.
Theory: measuring insecurity
A number of measures have been proposed, based on:
• General feelings of safety in the neighbourhood;
• The extent to which fear of crime influences daily habits;
• Worry about specific crime types;
• Frequency with which respondents feel fearful of crime or insecure
(and the situations in which this occurs).
Data: Five European victimisation surveys
Equesta de Seguretat Pública de Catalunya (Catalonia)
• Four sweeps between 2010 and 2013, average yearly sample size 14,000.
Crime Survey for England and Wales (formerly British Crime Survey)
• Five sweeps between 2010 and 2014, average yearly sample size around 40,000.
Cadre de Vie et Sécurité (France)
• Four sweeps between 2010 and 2013, average yearly sample size 15,000.
Crime prevention carousel (Budapest)
• One survey in 2003, 1,500 respondents.
Sicurezza del cittadini (Italy)
• 60000 respondents in 2008/09.
Analytical approach
• Dependent variable given by measure of victimisation or response
to certain insecurity question.
• For ordered responses relating to perceived insecurity: ordered
logistic regression.
• In the case of victimisation counts: negative binomial regression.
• In all cases, weighted to adjust for survey design.
• Two stage approach to model estimation:
1. Baseline models with consistent independent variables;
2. More comprehensive models with a larger range of variables.
• Baseline models consistent with fuller ones so only latter presented.
Independent variables
• Baseline demographic variables (included in all surveys):
–
–
–
–
–
–
Gender
Age
Unemployment
Student
Nationality different to country of survey
University degree
• Other variables (for inclusion, must be reported in at least 3 of 5 surveys):
– Live with partner
– Live in a detached house
– Number of people in household (sometimes separate counts of children/
adults)
– Years in the area
– Health (self-assessment)
– Victimisation of different crime types
Perceived insecurity: Neighbourhood safety
England and Wales:
“How safe do you feel walking alone in this area when dark?”
Catalonia:
“How safe do you feel in your neighbourhood?”
France:
“Do you personally feel unsafe in your neighbourhood or village?”
Hungary:
“How safe do you feel when you are alone in your local street during the
day?”
Italy:
“How safe do you feel when you walk alone at night in your
neighbourhood?”
Results: Neighbourhood safety
Neighbourhood safety
Female
Age
Unemployed
Student
Nationality outside country
Degree
Income
Live with partner
Detached house
Household size (children/adults)
Years in the area
Poor health
Violence
Vandalism
Burglary
Motor vehicle theft
Bike Theft
Robbery
Theft from the person
Harrassment and threats
EngWal
+
+
+
.
+
–
–
.
–
(./–)
+
+
+
+
+
+
+
+
+
France Catalonia Hungary
+
+
+
–
.
+
.
.
.
+
–
.
.
–
.
.
–
.
.
–
.
–
–
–
.
(./–)
+
.
+
+
+
+
.
.
+
+
.
+
.
+
.
.
+
.
+
+
.
Italy
+
+
+
+
.
–
+
–
+
+
+
+
.
+
+
.
+ More unsafe
– Less unsafe
Not significant
.
(p>.05)
Not included in
survey
Perceived insecurity: Influence on day to day lives
England and Wales:
“How much is your own quality of life affected by fear of crime?”
France:
“Do you personally feel preoccupied with the problem of
delinquency in your neighbourhood?”
Italy:
“How much does fear of crime influence your habits?”
Results: Influence on day to day lives
Influence on day to day lives EngWal
Female
Age
Unemployed
Student
Nationality outside country
Degree
Income
Live with partner
Detached house
Household size (child/adults)
Years in the area
Poor health
Violence
Vandalism
Burglary
Motor vehicle theft
Bike Theft
Robbery
Theft from the person
Harrassment and threats
+
–
.
.
+
.
.
+
–
(+/.)
+
+
+
+
+
+
+
.
+
France
Italy
+
+
.
.
+
+
.
.
–
(./.)
.
+
.
.
+
+
.
+
.
+
+
.
+
+
.
–
+
–
.
+
+
+
.
.
+
+
+ More influence
– Less influence
.
Not significant (p>.05)
Not included in survey
Victimisation: Burglary
Burglary
Female
Age
Unemployed
Student
Nationality outside country
Degree
Income
Live with partner
Detached house
Household size (child/adults)
Years in the area
Poor health
Violence
Vandalism
Burglary
Motor vehicle theft
Bike Theft
Robbery
Theft from the person
Harrassment and threats
EngWal
.
–
+
–
.
+
–
–
–
+/–
–
+
+
+
NA
+
+
+
+
France
.
.
+
.
+
.
+
.
+
+/.
.
.
+
+
NA
+
.
.
.
+
Catalonia
.
+
.
.
.
+
.
+
.
.
.
.
NA
+
+
+
.
Italy
.
–
.
.
.
.
.
+
+ Increased burglary risk
– Decreased burglary risk
. Not significant (p>.05)
Not included in survey
.
+
NA
+
+
.
.
.
Conclusions
• National victimisation surveys can enable international
comparisons, even when they do not have the same design
across countries.
• A number of cross-country differences exist when analysing
national victimisation surveys.
• In many cases, these differences appear to highlight actual
differences in the population, as opposed to differences
brought about merely by the question wording.
• Theories of vulnerability and victimisation and their relation to
perceived insecurity and victimisation are generally supported
by the data in five European countries.
Implications
• Future research could help identify the source of these differences.
Qualitative research likely to help.
• Victimisation surveys should consider cross-national comparisons as
a design criteria.
Thank you for listening
http://marginproject.eu
Download