sheppard g. kellam, m.d. air center for integrating education and prevention research in schools
DESCRIPTION
Young Adult Outcomes from the Good Behavior Game: a classroom behavior management program applied in 1 st and 2 nd Grades. Sheppard G. Kellam, M.D. AIR Center for Integrating Education and Prevention Research in Schools Presentation at SREE December 11, 2006 - PowerPoint PPT PresentationTRANSCRIPT
Young Adult Outcomes from the Good Behavior Game: a classroom behavior management program applied in 1st and 2nd Grades
Sheppard G. Kellam, M.D.AIR Center for Integrating Education and Prevention Research in Schools
Presentation at SREEDecember 11, 2006
Funding since 1984 by National Institutes of Health: NIDA, NIMH, NICHD
Co-Authors
Sheppard G. Kellam, M.D. 1,4
C. Hendricks Brown, Ph.D. 2
Jeanne Poduska, Sc.D. 1
Carla Ford, Ph.D. 1
Amy Windham, Ph.D. 1
Natalie L. Keegan 1
John Reid, Ph.D. 3
Nicholas Ialongo, Ph.D. 4
Hanno Petras, Ph.D. 6
Bonnie Copeland, Ph.D. 5
Linda Chinnia, 5
1 American Institutes for Research, Center for Integrating Education and Prevention Research in Schools2 University of South Florida, Prevention Science Methodology Group
3 Oregon Social Learning Center4 Johns Hopkins University Bloomberg School of Public Health
5 Baltimore City Public School System6University off Maryland
The Baltimore Education and Prevention Partnership
The Baltimore City Public School System (BCPSS) has collaborated in 3 generations of education and prevention epidemiologically-based randomized field trials.
Trials were directed at helping children master key social task demands in 1st and 2nd grade classrooms.
Interventions were tested separately, then together.
The 1st generation will be our main focus today, where the Good Behavior Game (GBG) was tested by itself and the children, now young adults, were recently followed to ages 19-21.
Developmental Epidemiological Randomized Preventive Trials
One of a set of current prevention research strategies
Intervention is directed at early risk factor to reduce risk and improve developmental paths
Defining population helps control selection bias
Allows study of means, but also variation in developmental paths and in impact—”who benefits and under what conditions”
Periodic follow-up to determine impact on paths and outcomes
Early Risk in Prevention and Education Research
Over the last four decades much has been learned about early risk factors and paths leading to behavioral, mental health, and school problems.
Most if not all are strongly related to academic failure, also a major risk factor for later school drop-out, delinquency, drug abuse, depression, and other problem outcomes.
Aggressive, disruptive behavior as early as 1st grade has been repeatedly found a risk factor for later school failure, delinquency, violence, drug abuse, and high risk behaviors.
Stancavage: Theory
Design of 1st Generation Baltimore Trial: Aimed at Aggressive Behavior and Poor Achievement Separately
41 1st grade classrooms in 19 schools
Across schools: 3 or 4 schools in each of 5 low to low/mid
SES urban areas were matched. 70% African American.
Schools were randomly assigned either to the standard
program (control); or to an enhanced curriculum (Mastery
Learning--ML); or to a classroom behavior management
program (Good Behavior Game--GBG).
Within intervention schools: Children were balanced across
all 1st grade classrooms. 1st grade classrooms and teachers
were randomly assigned to standard program classrooms or to
intervention classrooms.
Study Design cont.
In the 1st generation, the GBG trial was done over 1st and 2nd grades in 2 consecutive first grade cohorts.
1st cohort with 40 hours of teacher training and support thru the year. This was the effectiveness trial, and will be our focus today.
2nd cohort with same teachers with little training and support, tested the sustainability of results—the naïve trial.
Classroom Levels of Aggressive, Disruptive Behavior 8-10 Weeks After Random
Assignment in First Grade (control classrooms)
Level of Classroom Aggression
2.001.00
Fre
qu
ency
of
clas
sro
om
s
10
8
6
4
2
0
Mean = 1.85Std. Dev. = .42
N = 40
High Risk Children in Well vs Poorly-Managed Classrooms (control classrooms) If the top 25% of all children on aggressive
behavior were in disrupted classrooms, their risk of severe aggressive behavior problems by middle school was up to 59 times the average child’s.
If similar children were in well-managed classrooms, the risk was 2.7 times the average child’s.
Random assignment of children and teachers within schools allowed inferences re classroom effect.
Impact of Poorly Managed Classrooms on Teachers
The number-one reason for teacher burn-out is the inability to manage classrooms.
Teachers need tested tools to manage classrooms, i.e., to teach children how to be students.
A large portion of 1st grade teachers need such tools, e.g. ~50% in Baltimore.
Goals of the Good Behavior Game (GBG)
Provide teachers a classroom-wide method to socialize children into the role of student
Reduce classroom aggressive, disruptive behavior among children to enhance classroom teaching and learning
Prevent school failure, drug abuse, delinquency, and other problem outcomes
The History of the Good Behavior Game (GBG), a Classroom-wide Program
The GBG was originally developed by Barrish, Saunders, & Wolfe at the University of Kansas in the late 1960s
The GBG had been reported in over 20 scientific papers, none randomized field trials, prior to the Baltimore large scale epidemiologically based trials.
GBG Implementation
In Baltimore, the GBG consists of dividing the 1st grade class into 3 heterogeneous teams.
Teacher exhibits a large poster that states proper student behaviors, i.e., classroom rules.
Teams are rewarded for each child’s pro-social behavior, and not rewarded when a child is disruptive. It is “group contingent.”
GBG Implementation (cont’d)
Early in the year, the GBG was played
systematically for ten minutes, 3 times a
week, and the time extended over the
year.
Rewards were more abstract as the
year went on. It was carried out in first
and second grades.
Kellam: Framework
Measure of Early Classroom Aggressive, Disruptive Behavior
For this presentation a sub-set of Teacher’s Observations of Classroom Adaptation (TOCA R), a measure of each child’s social adaptation to classroom rules for student behavior
Structured 2 hour interview with the teacher, not a checklist
Ratings were obtained for each child in the classroom in fall and spring of 1st and 2nd grades, and thereafter in spring of 3rd through 7th grades
TOCA Aggressive, Disruptive Items on 6 point scales: (1) breaks rules, (2) harms others and property, (3) breaks things, (4) takes others property, (5) fights, (6) lies, (7) teases classmates (8) yells at others, (9) stubborn, (10) trouble accepting authority
The Young Adult Follow-up Data
Ages 19-21 2 Hour (average) telephone interview with
each respondent. Developmental, psychological and
psychiatric status, WHO version of the CIDI for diagnoses.
Juvenile court and school records. Social adaptational status in social fields
of parental family, school, work, intimate relations, marital family if any, peers.
GBG Impact vs All Controls on Any Service Use for Males
0
20
40
60
80
100
1 2 3 4 5 6
Pe
rce
nta
ge
of A
ny
Se
rvic
e U
se
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 74 )All Controls (n = 138 )
GBG Impact vs All Controls on Alcohol Abuse or Dependence Disorder for Males
0
20
40
60
80
1 2 3 4 5 6
Pe
rce
nta
ge
of A
lco
ho
l Ab
use
/De
pe
nd
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 73 )All Controls (n = 134 )
GBG Impact vs All Controls on Drug Abuse or Dependence Disorder for Males
0
20
40
60
80
1 2 3 4 5 6
Pe
rce
nta
ge
of D
rug
Ab
use
/De
pe
nd
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 72 )All Controls (n = 134 )
GBG Impact vs All Controls on Drug Abuse or Dependence Disorder for Females
0
20
40
60
80
1 2 3 4 5 6
Pe
rce
nta
ge
of D
rug
Ab
use
/De
pe
nd
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 103 )All Controls (n = 164 )
GBG Impact vs All Controls on Antisocial Personality Disorder for Males
0
20
40
60
80
1 2 3 4 5 6
Pe
rce
nta
ge
of A
SP
D
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 75 )All Controls (n = 138 )
GBG Impact vs All Controls on Regular Smoking for Males
0
20
40
60
80
1 2 3 4 5 6
Pe
rce
nta
ge
of R
eg
ula
r S
mo
ker
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 75 )All Controls (n = 136 )
GBG Impact vs All Controls on Completed High School for Males
0
20
40
60
80
100
1 2 3 4 5 6
Pe
rce
nta
ge
of C
om
ple
ted
HS
Teacher Ratings of Aggression: Fall of 1st Grade
GBG (n = 75 )All Controls (n = 137 )
GBG Effectiveness trial vs.
GBG Sustainability Trial Although the results from the 1st generation’s 1st
cohort were impressive, the results from the 2nd cohort-- the “sustainability trial”-- except for drug abuse and dependence disorder-- revealed less impact, but in the same direction.
Since the mid 1980s we have learned more about the problem of sustainability of practices and results over subsequent cohorts.
We are currently trying a model with BCPSS based on multi-level mentoring and continual monitoring.
Lessons Learned I
First-grade classrooms are of central importance to later academic, mental, and behavioral health.
A relatively simple universal method of classroom behavior management in 1st and 2nd grades, aimed at aggressive, disruptive behavior-- a risk factor shared by a set of long term outcomes-- appears to improve the set of long term outcomes.
Females are less responsive to GBG than are higher risk males. More research is needed re females.
Lessons Learned II
Without a system to mentor, model, and monitor teacher practices over time, GBG practices are prone to deteriorate.
Teachers need support from principals; principals from area leaders; area leaders from chief academic and executive officers.
Lessons Learned III
• Randomized field trials (RFTs) are vital in testing what works, for which children, in what conditions.
• Demographic epidemiology, analytic modeling, Pre-RFT observational studies—lead to testing with RFT designs
• Partnerships among researchers and school districts are essential to support such studies, and dissemination.
Thank you,
The End
Schools19
GBG6 schools
14 classrooms
ML7 schools
16 classrooms
External Control6 schools
11 classrooms
GBG Intervention8 classrooms238 students
GBG Control6 classrooms169 students
ML Intervention9 classrooms274 students
ML Control7 classrooms205 students
Control 11 classrooms310 students
23197%
14988%
25894%
19193%
25582%
18377%
12675%
18877%
15375%
22773%
17977%
11174%
17869%
14576%
19476%
Followed as young adult
Had baseline TOCA
2nd Generation Ed/Prev Trial in Baltimore Schools
Combined curriculum/instruction (C&I) and GBG from 1st trial
New family/classroom partnership program (FCP) tested separately
Children, teachers, intervention condition all randomly assigned within 9 schools
3 Classrooms randomly assigned within each school to: 1) C&I + GBG; 2) FCP alone; or 3) the standard program
Results: By middle school combined GBG and C&I improved both achievement and behavior
Family/classroom partnership by itself had modest impact
3rd Generation Ed/Prev Trial in Baltimore Schools Integrate 3 components into 1 Whole Day
Program (WD): GBG+ C&I+ F/C partnership
8 Development Schools helped design and refine interventions and measures
Within 12 trial schools, random assignment of all 1st grade children, teachers, and classrooms to WD or standard program
Children in 12 Whole Day 1st grade classrooms are compared to children in 12 Standard Program classrooms
Efficacy
Effectiveness
Sustainability
Going to Scale
Sustaining Systemwide
Efficacy
Effectiveness
Sustainability
Going-to-Scale
Sustaining
Systemwide
Phases of Education and Prevention Trials
Baltimore analytic model
ClassroomBehavior
Management Classrooms
DecreasedAggressive, Disruptive
Classrooms
Teachers’ EffectiveAcademic
Instruction
PoorReading Skills
ImprovedReading Skills
Decreased Later Conduct & Anti-
social PersonalityDisorders
Effective Family-Classroom
Partnerships
•••
Whole DayWhole Day First Grade Education and
Prevention Program
Decreased Later DepressiveDisorders
Depressive Symptoms
DecreasedDepressive Symptoms
PoorAchievement
IncreasedAchievement
Decreased Later Substance Abuse
Individual Aggression
DecreasedIndividual Aggression
School Success & Decreased
Drop-Out
Other mediating or moderating variables:
•Family and poverty
•Deviant peers
•School building
•Community economic health, resources, drugs, and violence
Kellam: Framework
DEVELOPMENTALEPIDEMIOLOGY:
directed at early proximal targets
MORE IMMEDIATE RISK:
directed at more recent
proximal targets
COMMUNITY PREVENTION:
directed at Community & School proximal
targets
COMMUNITY / SOCIETAL:
directed at Policies & Laws as
proximal targets
Prevention Research Strategies