Observational constraints on mergers creating magnetism in massive stars
Pauwels -Wikström Sat test flemish society of criminology (2011)
1. How invariant is the interaction
effect between exposure and
propensity in the explanation of
offending?
Lieven J.R. Pauwels (Ghent University)
Per. Olof H. Wikstrom (Cambridge
University)
University of Cambridge
2. Introduction
• Central problems in the study of crime
causation
• SAT as theoretical framework
• Comparison of results
• Conclusion and discussion
2Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
4. Male
Live in a
split family
Immigrant
Have a difficult
temperament
Have been subjected
to poor parenting
Have been
physically or
sexually
abusedHave a
low
heart rate
Have delinquent
friends
Have been
bullied in
Are unemployed
The risk of
punishmen
t
is low
Grow up in a
disadvantaged
neighbourhood
Have
learning
difficulties
Played too much
violent video
games
Abuse
alcohol
or drugs
Spend too much time
in risk environments
Live in a morally
declined society
Are socially
excluded
Have low levels
of serotonin
Do not find the
law legitimate
Live in a
consumerist
society
Mentally ill
ACT OF
CRIME
DOES EVERYTHING
MATTER ?
DOES ANYTHING
MATTER?
4Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
5. • The study of deviance and crime has traditionally been characterized by a multitude of seemingly
unrelated and competitive theories” (Liska, Krohn & Messner, 1989)
• No simple theory in the crime/deviance area /.../ has proven to be more than minimally satisfactory
in overal explanatory ability, in applicability to a wide range of types of deviance, or in empirical
support for its tenets. All are plausible, yet they fail as general theories. (Tittle, 1995)
• Criminology has failed to make scientific progress in the past 20 years in the sense of falsifying
some theories and accumulating verified knowledge in the context of other theories(Bernard,
1990)
• Criminology risk being a field of study in which many ideas are developed and all are chosen – in
which all theories have equal claim to legitimacy and in which only the most highly specialized
scholars can separate the theoretical wheat from the chaff. (Cullen, Wright & Blevins, 2008)
5Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
6. Key analytical problems in criminological theoryCriminological theory generally fails to fully address:
(i) what crime is (to clearly define what it is the theory
aims to explain)
(ii) what it is that moves people to engage in acts of crime
(to present an adequate action theory)
(iii) How personal and environmental factors interact
in moving people to engage in acts of crime (to properly
integrate key insights from personal and environmental
explanatory approaches)
(iv) The role of the broader social conditions and individual
development / life-histories (to analyse their influence
not as causes but as causes of the causes)
6Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
7. 1. A cause has to be a cause of something
(criminological theory do not always clearly specify what it is
it aims to explain)
2. A cause is something that produces something else
(a cause is something that has powers to initiate (help initiating)
the process (mechanism) that produces the effect)
3. Human action (incl acts of crime) is an outcome of
the causal interaction between person and environment
(there is a need for a situational model that explains how person
and environment factors interact in moving people to engage in
acts of crime)
4. There are causes and there are causes of the causes
(criminological theory often confuses causes and causes
of the causes)
7Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
8. The problem of
correlation and causation
“…it is difficult to see why anyone would be interested
in statistical association if the findings were not in
some way relevant to the understanding of causative
mechanisms” (Rutter, 2009)
“…statistical correlation explains nothing;
it is what cries for explanatory models”
(Bunge, 2006)
“Statistical techniques, no matter how powerful in revealing
social regularities, cannot at the same time be used to crank
out causal explanations of the regularities”
(Goldthorpe, 2000)
8Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
9. Key hypothesis derived from SAT:
SAT (Situational Action Theory):
- Peoples’ crime involvement dependent on their crime
propensity and their criminogenic exposure
- Interaction between crime propensity and exposure
- The importance of criminogenic exposure increases with
a person’s crime propensity
- Specifically for those with low crime propensity,
criminogenic exposure is practically irrelevant
9Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
10. Confounding causes with
correlates
• Immigrant background and gender are
correlates, not causes of offending!
• Too many studies include correlates as
“statistical controls” – Why?
• “The philosopher will be surprised to see that
/…/ investigators take properties („variables‟)
such as age and sex to be possible causes”
(Bunge, 2001:70).
10Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
11. Key Hypothesis
Preliminary COMPARITIVE TEST of SAT:
Does the interactive hypothesis (offending =
exposure * propensity) stated in SAT hold in sub
groups by gender and immigrant background in both
data?
- PADS+ study (Panel study-Peterborough) (averages
2003-2007 samples)
- Antwerp Young Adolescents Survey (cross-
sectional school, 23 schools in 42 neighbourhoods,
2005)
11Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
12. Measures
• Propensity:
• B: combined index of delinquency
tolerance and low self-control
• UK: combined index of delinquency
tolerance, low self-control, shame and guilt
• Both represent propensity
12Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
15. Analysis Plan
• OLS-regression main effects and
interaction term (despite assumptions…)
• Comparing means in sub groups (tertiles
and +/- 1 std)
• Not: multiple group comparison
• Basic analysis of patterns in all sub groups
15Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
16. Antwerp data: girls only
16Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
17. Antwerp data: boys only
17Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
18. Peterborough data: boys only
Exposure to criminogenic moral settings
1.00.00-1.00
Meantotaloffending
140.00
120.00
100.00
80.00
60.00
40.00
20.00
0.00
British boys
high
medium
low
Propensity
Expsoure to criminogenic moral settings
1.00.00-1.00
Meantotaloffending
250.00
200.00
150.00
100.00
50.00
0.00
Immigrant boys
high
medium
low
Propensity
18Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
19. Peterborough data: girls only
Exposure to criminogenic moral settings
1.00.00-1.00
Meantotaloffending
60.00
50.00
40.00
30.00
20.00
10.00
0.00
British girls
high
medium
low
Propensity
Exposure to criminogenic moral settings
1.00.00-1.00
Meantotaloffending
40.00
30.00
20.00
10.00
0.00
Immigrant girls
high
medium
low
Propensity
19Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
20. Peterborough data Antwerp data
Immigrant boys Non-immigrant
boys
Immigrant boys Non-immigrant
boys
B/beta
(SE)
B/beta
(SE)
B/beta
(SE)
B/beta
(SE)
Propensity to
offend
9.31/0.39***
(2.334)
8.81/0.35***
1.773
0.126/0.320***
0.014
0.129/0.416***
0.012
Exposure to
criminogenic
moral settings
4.67/0.08ns
6.280
0.76/0.00 ns
4.07
0.831/0.266***
0.121
0.547/0.203***
0.108
Interaction term 8.15/0.57***
1.115
2.93/0.30***
(0.671)
0.062/0.222***
0.010
0.073/0.265***
0.010
R²*100 66.7 29.0 42.0 41.01
N 65 261 599 506
TABLE 1: OLS-regression of immigrant and non-immigrant boys in Peterborough and Antwerp
*: p<0,05, **: p<0,01, ***: p<0,001
20Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
21. Peterborough data Antwerp data
Immigrant girls Non-immigrant
girls
Immigrant girls Non-immigrant
girls
B/beta
(SE)
B/beta
(SE)
B/beta
(SE)
B/beta
(SE)
Propensity to
offend
1.67/0.29***
0.685
3.90/0.48***
(0.562)
0.109/0.425***
0.009
.098/0.402***
0.009
Exposure to
criminogenic
moral settings
3.94/0.36**
(1.295)
1.10/0.07ns
(0.191)***
0.611/0.253***
0.085
0.663/0.303***
0.008
Interaction term 0.77/0.31***
(0.196)
0.94/0.25***
(0.191)
0.051/0.210
0.008***
0.068/0.274***
0.008
R²*100 51.0 45.4 41.7 40.7
N 80 254 631 528
TABLE 2: OLS-regression of immigrant and non-immigrant girls in Peterborough and Antwerp
*: p<0,05, **: p<0,01, ***: p<0,001
21Pauwels & Wisktröm 3de
criminologisch Forum VUB 11
22. Conclusions and discussion
Crime propensity and criminogenic exposure
are strong predictors of adolescents’ crime involvement
There is a significant interaction effect between crime propensity
and criminogenic exposure
(exposure has a stronger effect for persons with stronger propensity)
These findings holds for both immigrant and non-immigrant boys and
girls and for both cities
These findings thus support a key basic proposition of
Situational Action Theory
Still: differences in R square and problems of measurement
invariance in surveys?
Future comparative studies: How invariant are the causes of the
causes?
22Pauwels & Wisktröm 3de
criminologisch Forum VUB 11