Youth Delinquency and the Neighbourhood Context

1
Youth Delinquency and the Neighbourhood Context

• Susan McVie and Paul Norris
• Edinburgh Study of Youth Transitions and Crime
• University of Edinburgh

2
Background

• The relationship between crime and place has a
  long established history, with crime being more
  prevalent in disadvantaged and disordered
  neighbourhoods.
• Theories have focused on physical conditions of
  the area, socio-structural characteristics of
  residents and capacity of residents to
  self-regulate behaviour.
• Recent advances in computing software have
  improved capacity for multi-level modelling,
  allowing contextual effects at different levels
  to be measured.
• Gaps in theory about mechanisms at work,
  relationships between individual and
  neighbourhood level influences and differential
  impact of neighbourhoods on other problematic
  behaviours (e.g. drug use).
• Need to inform policy focus on community based
  responses to crime.

3
The Edinburgh Study
Aims to expand current understanding of youth
offending by exploring patterns of offending
within the context of individual development,
agency intervention and neighbourhood effects.
Prospective longitudinal study of a single age
cohort including all children starting secondary
schools in city of Edinburgh in August 1998
(n4300). Complex research methodology involving
multiple sources of information over several
annual sweeps (6 to date). High response rates
(81 at sweep 6). Design includes a geographic
information system through which 91 Edinburgh
neighbourhoods have been defined. Various types
of geo-coded data (police recorded crime, census
data and community survey) are linked to cohort
members via home postcode.
4
Edinburgh neighbourhoods

• 91 neighbourhoods defined mainly to ensure
  statistical power in terms of population size
  (approx 4,500-5,000 in each) and geographically
  contiguous tracts internally homogenous on key
  census indicators.
• Index of social and economic stress (demographic,
  household, housing and economic instability).

5
Neighbourhood effects on youth delinquency and
drug use
6
Patterns within neighbourhoods

• There is a long history of research that shows
  large differences in offending rates between
  different types of neighbourhood.

• Much less research has focused on the impact of
  neighbourhood characteristics on patterns of drug
  use.

7
Variables of interest

• Ordinal measure of delinquency
• Based on frequency of involvement in 17 forms of
  delinquency in the last year (72 prevalence).
• Highly skewed measure grouped into 6 ordinal
  categories (0, 1-5, 6-10, 11-15, 16-20, 21)
• Ordinal measure of cannabis use
• Based on frequency of cannabis use in last year
  (29 prevalence).
• Also highly skewed with many zeros, so grouped
  into five categories (0, 1, 2-3, 4-10, 11)
• Ordinal measure of hard drug use
• Based on frequency of use of six types of class A
  or B drugs in the last year (7 prevalence).
• Even more skewed, grouped into 5 categories (0,
  1, 2-3, 4-10, 11)
• All measured at sweep 5 (age 16)

8
Individual level explanatory factors

• A wide range of individual characteristics have
  been found to be linked to youth delinquency and
  drug use.

• This analysis includes only a small number of
  individual control variables
• Individual gender
• Individual impulsivity
• Family structure
• Family socio-economic status

9
Individual level correlations
Correlations showing the strength of relationship
between the individual explanatory factors and
the three dependent variables.
10
Area level explanatory factors

• 2001 Census measures
• - economic deprivation
• - population instability
• 2001 Police recorded crime data
• - crime categories representing street crime
  (vandalism, fire-raising, housebreaking, vehicle
  crime and non-sexual violent crime)
• 2001 Community survey (n1642)
• - neighbourhood dissatisfaction
• - community safety
• - physical disorder
• - collective efficacy

11
Area level correlations
Volume measures of delinquency and drug use were
aggregated to the neighbourhood level, and
Spearman correlation used to measure strength of
association between these and the neighbourhood
level variables (n91).
Notes Significant at the 95 level or above.
12
Combining area and individual data

• Single level analysis breaks assumptions of
  statistical methods.

• Single level analysis may miss interesting causal
  relationships.

13
Ordinal regression model
Notes Individuals were clustered according to
residential neighbourhood. All variables with
coefficients listed were significant at the plt.05
level. Cannabis user added for hard drug use
model only.
14
Conclusions

• Neighbourhoods have some impact on problematic
  behaviour, although in cross-sectional analysis
  the impact of individual factors is greater.
• Different neighbourhood level factors explain
  delinquency, cannabis use and hard drug use,
  suggesting that different theoretical frameworks
  and policy responses are required.
• Findings support policies that aim to reduce
  crime by focusing on structural inequalities and
  economic deprivation within communities.
  However, such policies are unlikely to have any
  impact on cannabis use with is associated
  affluence at both the neighbourhood and the
  individual level.
• Strategies for tackling hard drug use might be
  focused in high crime areas, but more needs to be
  understood about this type of substance use in
  teenagers.
• Simple analysis can be misleading (correlations
  showed delinquency more strongly associated with
  neighbourhood factors than drug use).

15
Future analysis

• More advanced multi-level modelling to examine
  the precise amount of variation explained by
  neighbourhoods, and whether the effect of
  significant individual and neighbourhood factors
  varies by neighbourhood.
• Expand the pool of potential explanatory factors,
  including family, school, peer-related factors,
  etc to more fully understand the relationship
  between place and behaviour.
• Explore the inter-relationship between different
  contexts, such as neighbourhoods and formal
  interventions by the justice system, and the
  differential impact of perceptions and realities.
• More detailed investigation of longitudinal
  causal mechanisms by exploring the effect of
  neighbourhood factors on trajectories of
  offending and substance using behaviour (started
  work in the area on property crime).

16
Further information

• www.law.ed.ac.uk/cls/esytc
• Edinburgh.study_at_ed.ac.uk