Improving social-cognitive skills among disadvantaged
Youth
1
Harold Pollack, Co-director, University of Chicago Crime Lab
COMPREHENSIVE SOLUTIONS FOR CRIME AND VIOLENCE PREVENTION
26 – 29 de Junio, 2013Cali, Colombia
One face of Chicago’s Youth Violence problem
• Chicago’s 506 homicide led the nation in 2012.• Hadiya Pendleton shooting right after she
performed at President Obama’s second inaugural one of many prominent homicides in which the victims and/or perpetrators were Chicago youth.
Plenty of other cases & faces here in Chicago
• Chicago had 506 homicides in 2012, the most in the United States, up from 436 in 2011.– 10-year-old Nequiel Fowler shot and killed after being hit
by stray bullet. She was kneeling to tie the shoe of her blind little sister.
– January 17, 2013, Tyrone Lawson, 17, was shot and killed after a high school basketball game. The game ended with an altercation between the teams while lining up to shake hands. But tensions spilled out into the parking lot and Lawson was shot and killed for no apparent reason.
– Maybe more typical: Saturday June 2, 2012, two groups of teens arguing in the street about whether someone stole a bike. As two groups start to separate, someone pulls out a handgun and fires into other group, hits Jamal Lockett, age 16, in the chest, who dies at at Northwestern Memorial Hospital.
CHICAGOJUNE 2, 2012
• Male: 90% of all homicide victims & offenders (2011 CPD)
• Young: 53% homicide victims & 57% offenders < age 25• 79% victims & 83% offenders < age 35
• Guns: 83% of homicide victims shot• Public: 77% of homicide victims found outdoors• Impulsive: 73% homicides attributed to “altercation” • Only ~10% to gang disputes over narcotics. • Jamal Lockett’s older sister Adora Dismukes: “They just did this for
nothing, killed my little brother for nothing. Nothing. It’s just pathetic.”
• One-third of victims high blood alcohol.
Common features of violence in Chicago(not just Chicago)
CHICAGO HOMICIDES HAVE DECLINED, NOT AS FAST AS IN TWO KEY PEER CITIES.
Arguments across political spectrum that violence is very deeply- rooted
• Liberal and conservative arguments for pessimism regarding incremental progress.
– Liberal argument.– Violence fundamental outgrowth of economic inequality,
blocked opportunities, segregation, and discrimination.– Conservative version.– Violence fundamental outgrowth of adverse cultural trends
including family breakdown, adverse media messages, and more.
– Both perspectives have some merit. Both also have serious limitations as either guides to policy or as a lens to predict variation over places and over time.
COMMITTED CRIMINALS
COMMITTED CRIMINALS
U.S. incarceration rate, 1925-2008prisoners per 100,000 population
A DIFFERENT VIEW OF WHY PEOPLE COMMIT CRIME
Young men + disagreement + impulsivity + gun = dead body
Fundamental equation of many homicides
Policy response to this equation
• Implicit logic model behind public policy:• You are a pissed-off 17 year old boy surrounded by your
friends• You are susceptible to sensation seeking & peer influences
(brain changes starting in early adolescence), myopic decision-making, “catastrophizing” (make negative events even more negative), low impulse control / self-regulation, & “hostile attribution bias”
• “If you pull that 9mm out of your waist band something bad will happen – or at least it might, with considerable delay”
• One response:• If not enough youth respond to that threat, let’s add 2 more
years onto that 8 year prison sentence…
BLUE
• Could instead try to improve decision-making capacity• That is, remediate deficits in “social-cognitive” skills [such
as] impulse control, anger suppression, future orientation• These are the strongest predictors of recidivism risk among juvenile
offenders (Monahan, Steinberg, Cauffman, Mulvey, 2009 Dev Psych)• More important than measures of consideration of others, sense of
personal responsibility, resistance to peer influence• Large & growing body of research shows social-cognitive skills
correlated with schooling, earnings, other outcomes (e.g., Moffitt, Heckman)
• What we haven’t known is whether it is really possible to modify these skills – particularly among at-risk adolescents• Not a surplus of successful interventions for this population in general
Rather than change long-term incentives facing a youth who is not in good decision-making frame of mind…
OUR STUDY
• Held design competition in 2009• Selected social-cognitive skill development program
(CBT) by two local non-profits • Youth Guidance & World Sport Chicago
• Raised $1 million from NICHD, U of C provost and SSA, Chicago Community Trust, Exelon, and several local foundations (Joyce, MacArthur, McCormick, Polk, Spencer) and NORC Population Research Center
• One of the largest experimental test of social-cognitive skill developments in realistic conditions in Chicago Public Schools
• “Becoming a Man” - application of CBT, mentoring, & positive youth development targeted at urban adolescent males• 27 week curriculum of group sessions (1 session / wk), 10-15 students / group• Sessions occur during the school day
• Didactics, role-playing, homework, group activities• Impulse & anger management (11/27 sessions), future orientation (8/27), social information processing (8/27), moral reasoning (13/27) •(Ex) the Fist exercise
•Non-traditional/combat sports• Archery (!), boxing, judo, karate, etc.• Coaches also trained in BAM principles• Develop /reinforce social-cognitive skills• Increase program participation
21
INTERVENTION IMPLEMENTED IN CHICAGO PUBLIC SCHOOLS IN 2009-10 AY
22
Intervention SchoolsAustin Polytech (High School)Banneker (Elementary)Bass (Elementary)Clemente (High School)Crane (High School)Douglass (High School)Fenger (High School)Harper (High School)Jordan (Elementary)Juarez (High School)Little Village (Elementary)Orr (High School)Parker (Elementary)Robeson (High School)Yale (Elementary)
ADMINISTRATIVE DATA USED FOR ALL OUTCOMES
• Student-level Chicago Public School (CPS) records• Enrollment / dropout status, grade of enrollment, course
grades, achievement test scores, absences• CPS staff matched data from within internal
administrative records • 2008-9 AY (baseline controls)• 2009-10AY (program year outcomes), 2010-11 longer-
term outcomes• Illinois State Police (ISP) arrest records• Complete census of all arrests made in IL, now has good
coverage of juvenile as well as adult arrests• Linked together with biometrics (fingerprints)• Date of arrest, criminal charge(s) filed (we divide up into
violent, property, drug, motor vehicle, and other crimes)
Analysis Strategy
Intent to Treat (ITT)• Average difference between those assigned to
treatment vs control• We account for student baseline characteristics to
more-precisely estimate program effects
Examining impact of interventions on kids who actually participated (LATE).
• Assumes assignment to treatment group only affects outcomes through participation (exclusion restriction)
• Use control complier mean (CCM) to assess relative size of impact
OUTCOME MEASURES
• Form unweighted averages of components in z-score form, standardized on control group (following Kling et al. 2007)• Academic composite• GPA, # days present, indicator for “still in school”• In following year, we examine school transfers,
attendance in juvenile justice schools, as well.• Arrests• Look at violent, property, drug, and other crime
arrests separately (may be affected differentially). We also show aggregate social costs as summary measure.
DESCRIPTIVE STATISTICS
Note: One school excluded from sample due to randomization failure
RECORDED PROGRAM PARTICIPATION
SUMMARY BOTTOM LINE SLIDE
CRIME RESULTS
CHANGES IN WHICH SCHOOL ATTENDED
BENEFIT-COST ANALYSIS• Can compare social benefits to costs of program
($1,100 per participant) to assess whether results are large or small
• Crime benefits• Given fade out, can be fairly certain these benefits have
already accrued• Difficult to assign dollar values to crime, depends on value
placed on avoiding a homicide, intangible costs of crime• We use range of approaches, present low and high
estimates• Schooling benefits• Most studies estimate benefit of graduation, but our
students mostly too young to have graduated• We extrapolate graduation effect from correlation
connection between grades and diplomas. Much more tenuous, but big enough to be important even if we drastically over-estimate graduation changes
WHAT HAVE WE LEARNED?
• Large-scale “effectiveness” of social-cognitive skill development in challenging field conditions• Using estimates for social cost of crime,
benefit/cost ratio between 2 and 9:1 (even ignoring schooling benefits)• Despite fade-out, targeting program during a
period of high-risk behavior (adolescence) means even temporary effects have high social payoffs• Not a panacea, but a potentially cost-effective
part of a portfolio of solutions
Next steps
• Adding tutoring/small group teaching arm to enrich academic skills
• Early pilot results promising.• Perhaps our next talk.
Reminder (after Roseanna Ander’s talk)• Value of broadening focus of institutions that
work with at-risk youth– Social-cognitive skills, not just academic &
vocational skills• Multiple incremental changes vs. one “home
run”– There will never be a single Salk vaccine for violence– Goal: optimal portfolio of interventions
• Under-investment in some other things?– We do a poor job right now of identifying & helping
the 8th grade boy who has (say) missed 8 weeks of school
– This needs to be a bigger piece of the mix.
REAL EVIDENCE ATTRACTS ATTENTION
(Thank goodness I brought my new camera.)