The role of adversity in shaping
children’s self-regulation:
Drawing from experimental and nonexperimental evidence
C. Cybele Raver
Professor
Dept. of Applied Psychology
Director, Institute of Human Development and Social Change
My Programs of Research
•
•
•
Measurement and models of poverty
– Raver, Gershoff, & Aber (2007). Testing equivalence of mediating models of income,
parenting, and school readiness for White, Black, and Hispanic children in a national
sample. Child Development
Children’s emotional and behavioral regulation as a predictor of school readiness
– Raver (2004). Placing emotional self- regulation in sociocultural and socioeconomic
contexts. Child Development.
– Ursache, Blair, & Raver, (in press). The promotion of self-regulation as a means of
enhancing school readiness and early achievement in children at risk for school
failure. Child Development Perspectives.
Adults’ mental health and caregiving
– Raver (2003). Does work pay?: The effects of employment on depressive symptoms
and parenting among low-income families. Child Development.
– Zhai, Raver, & Li-Grining, (2011). Classroom-based interventions and teachers’
perceived job stressors and confidence: Evidence from a randomized trial in Head
Start settings. Early Childhood Research Quarterly.
Placing developmental models in context
Theory
Employment
Psychological
wellbeing
Poverty
Caregiving,
instruction
Child selfregulation
Welfare
Reform
Emotional
climate
Policy & Prevention
“bottom line”
Social & academic
competence
HS
School
Overview
• Self-regulation in the context of poverty
• “Whirlwind tour” of the Chicago School
Readiness Project (CSRP)
• RCTs versus the ‘real world’ – lessons from
longitudinal follow-up of CSRP
• Implications for developmental science,
prevention science, and for policy
Child and Family Poverty in the U.S.
In 2010, 18.8%
of Pennsylvania
children lived <=
poverty line, up
from 15.9% in
2007.
Today, national
child poverty
rate = 21.2%.
DeNavas-Walt, C., Proctor, B. D., & Smith, J. C. (2010). Income, Poverty, and Health Insurance Coverage in the United States: 2009.
U.S. Census Bureau, Current Population Reports, P60-238,.Washington, DC.: U.S. Government Printing Office. Available at:
http://www.census.gov/prod/2010pubs/p60-238.pdf
Educational outcomes associated with child
poverty
• Poverty associated with children’s lower academic achievement, higher
grade retention, and higher risk for dropping out.
– Example: In Chicago, 86.50% of children eligible for free/reduced-priced
lunch.
– More than 25% of Chicago’s 24,000 third graders were not able to pass
their school district’s gated proficiency standards, and subsequently faced
high levels of long-term educational risk (Roderick & Nagaoka, 2005).
• Low-income children face higher risk of behavioral problems, lower levels
of competence in getting along with teachers and peers (Bolger et al, 1995)
• Gap at school entry: Reading scores for kindergartners from lowest SES
group are 60% lower than those of students at the highest end (Burkham &
Lee, 2002).
Is the solution to ask teachers to “teach
more?”
• Cumulative poverty-related stressors (e.g. instability,
family/community violence) -> children’s higher risk of
withdrawal, disruptive behavior.
• Under-resourced classrooms in low-income communities are
ill-equipped to cope with the emotional and behavioral needs
of children facing high levels of risk (Dodge et al, 1999)
– 16% of children in Head Start classrooms exhibited disruptive/unsafe behaviors,
e.g. kicking, hitting and threatening at least once a day (Kuperschmidt, et al 2000).
– Children’s disruptive behavior may limit not only their own but others’
opportunities for learning
“Lenses” of developmental research
Definitions
• Emotional and behavioral self-regulation – temperament research
– emotional “reactivity and regulation” + effortful/inhibitory control
(Thompson, 1994; Eisenberg & Fabes, 1992)
• Executive function – neurocognitive research
– children’s working memory, attention deployment, and ability to
inhibit prepotent impulses in order to meet external demands
(Diamond & Taylor, 1996; Blair, 2002).
Preschool– Children expected to modulate distress in context of
highly arousing and potentially stressful situations (e.g. conflict over
toys, teacher expectation of compliance).
If interventions can be designed to support children’s self-regulation,
will that improve their chances of school success?
Evidence that children’s self-regulation is
influenced by environment, malleable?
EF and poverty as example:
•
Previous findings of SES gradient and EF
(Hackman & Farah, 2010).
•
Strong support for model of experiential
canalization, where poverty-rel. stress->
greater allostatic load ->alterations in
HPA axis ->alterations in PFC functioning
(Blair, 2010; Blair, Raver et al., 2011)
•
Recent work suggests may be
particularly important to take into
account chronicity of poverty, over time
(Duncan & Magnuson, 2009; Raver, Blair,
Willoughby & FLP Key investigators)
Raver, Blair, Willoughby, & FLP Key Investigators (to be revised and
resubmitted)
Asking two questions:
1) Is self-regulation “malleable?” and
2) Does self-regulation “matter” for academic
achievement?
– Problems with inferring that low-resource
environments shape children’s difficulties with
self-regulation
– Problems with inferring children’s emotional and
behavioral problems as cause of academic difficulty:
Perhaps children act out because they’re frustrated
they can’t read.
SR
Inc
Ach
X3
– Solution: Random assignment of some preschool
sites to receipt of “treatment” targeting children’s
emotional and behavioral regulation and other sites
to “control group”
• Allows for empirical tests of causal claims.
Integrative theoretical model of the role of children’s
emotional and behavioral adjustment in predicting school
readiness (Raver & Jones, 2003)
Fall/ Sc hool entry
PRE-K S CHOOL Y EA R
Spring/Trans ition to Kindergarten
Teac her
instruction
(language/
literacy)
c hild
language/
pre-literacy
T2
child
language/
pre-literacy
T1
child
emotiona l/
behav ioral
adjustment
T1
School
Readines s
T3
(e.g. Fall
Kindergarten)
Teac her
classroom
manage ment
(provision of
regulatory
support)
Cu mulative
f amily/
neighborhood
risk
c hild
emotiona l/
behav ioral
adjustment
T2
Acknowledgements
PI: C. Cybele Raver
Co-PI. Stephanie M. Jones,
Co-Is: Christine Li-Grining, Fuhua
Zhai
Research Staff:
Amy Lowenstein
Julie Neuspeil
Rachel Pess
Nicole Tirado-Strayer
Emily Pressler
Kathleen Zadzora
Genevieve Okada
Dana Charles
Allison Friedman
Breeze Luetke-Stahlman,
Ta-Tanisha Young,
Tiffany Hayes,
Latriese Sardin Adjei,
Molly Metzger,
Bonnie Solomon
Chevon Townsend
Radiah Smith-Donald
Kina Smallwood
+ 20 grad student assessors
Support:
– W.T. Grant Foundation
– Interagency Consortium on
School Readiness, (NICHD)
– Spencer Foundation
– McCormick Tribune Foundation
Intervention staff:
Sybil Madison-Boyd
Darlene Jones-Lewis,
Kimya Barden
Elysia Aufmuth
Greg O’Donnell (Metro
Family Svcs)
Partnering Delegate Agencies:
Department of Children Youth Services
Chicago Public Schools
Ounce of Prevention/Casa Central
Hull House Humbolt Park
Boys & Girls Club
Carole Robertson Early Learning Centers
CYC Rebecca Crown and Ida Mae
Fletcher Centers
South Shore United Methodist Center
YMCA
Hypotheses
• The principal aim of CSRP is to improve low-income preschool-aged
children’s school readiness.
– Short-run: Test whether children in treatment-assigned classrooms
show
• significantly improved behavioral self-regulation, and
• significantly greater gains in pre-academic skills
than do children in control-assigned classrooms, in spring of Head
Start year.
Long-run: Does additional investment in preschool make
substantive difference in young children’s chances of school
success?
• If accrued across treatment year, are gains sustained as children
make transition into Chicago Public Schools?
Overview of CSRP services
Fall
Winter
Spring
Teacher training + coaching
Stress reduction
1-on-1 for children with
highest EBPs
Train teachers to respond in proactive ways to support children’s regulation
and to limit disruptive behavior.(Webser-Stratton et al, 2001; Brotman et al, 2005, Kellam, et al,
1998)
Paired teacher-training with “coaching” by MHC in light of
a) concern for dosage
b) need for stress reduction
Direct “1-on-1” services by MHC. Children’s behavior problems - beyond
the scope of teachers’ responsibility.
Selecting 18 sites (35 classrooms):
Balancing generalizability with feasibility
•
•
Exclusion criteria
Neighborhoods < 400 HS-eligible
children
•
Demolition of CHA housing &
gentrification –
•
Communities experiencing < median
levels of crime.
•
18 Sites, 90 teachers, 602 children
enrolled in two cohorts
•
Rates of parental consent for child
participation ranged from 66.6% to
100% across all sites, Mean = 91%, SD
= 6% .
Sample
•
66% of families identified as African American/Black and the
remaining families identified as Hispanic/Latina/o
•
Families enrolled in CSRP reported monthly incomes of $1,168 on
average (average income-to-needs = .67), with 67% supporting their
families as single heads of household.
•
Most parents were working, and many faced a range of serious
poverty-related stressors.
•
Teachers were also supporting families on low incomes and 1/3rd of
teachers reported feelings of high levels of stress and low
confidence in managing children’s behaviors in the classroom.
•
Classroom quality at baseline was “adequate,” on average (Mean
ECERS = 4.97), with substantial variability (SD = 0.77).
Design
Each Head Start site matched with another “sister” site on range of
demographic characteristics of families, site characteristics indicating program
capacity, etc. and then randomized to tx vs. control status.
Treatment – “Package A”
Control – “Package B”
Services
provided
30 hours of training offered.
MHCs with a MSW spent 1 day a week
in classroom, from Sept to May of
school year.
Teacher’s assistant
(1 day1/wk)
Sites
9
9
Classrm
18
18-1
NB: Originally funded for 10 sites– “hopelessly underpowered” (Cook, 2005). Boosted to 18
(expecting attrition of 1 pair) and included covariates. Best guess that ICCs were <.05
Data Collection
•
•
•
Site level
– 14 characteristics (e.g. % of teachers with BA) from PIR
Classroom level
– CLASS (La Paro & Pianta, 2004) (Sept, Jan, Mar, May) Classroom
climate, teacher sensitivity, etc.
– Teacher psychosocial characteristics (K-6, job stressors)
– Classroom-level Covariates
• Classroom quality, ECERS, structure
Child level
– Family interview (Fall baseline, Fall 1-year-followup)
– Teacher report of internalizing and externalizing behavior problems
(Sept, Jan, March, May)
– Direct assessment (Sept, May)
– Kindergarten follow up (teacher report, school record).
CSRP: Significant reductions in children’s externalizing and internalizing
behavior problems
Overall
Gender
Race
≥2 risks
One risk
No risk
Other
Hispanic
Black
Girl
Boy
CSRP Treatment Effects
Figure 2. Effects of CSRP Treatment on Child BPI Externalizing Scores
1.0
0.0
-1.0
-2.0
-3.0
-4.0
-5.0
-6.0
-7.0
-8.0
-9.0
-10.0
-11.0
-12.0
-13.0
Family Risks
Raver, Jones, Li-Grining, Zhai, Metzger & Solomon (2010). Journal of Consulting and
Clinical Psychology
The Preschool Self-Regulation Assessment (PSRA)
Smith-Donald, Raver, Hayes & Richardson, 2007; Li-Grining & Raver (revise & resubmit) ; Raver et al (in press)
We embedded direct assessment of
children’s self-regulation within a
standardized battery that included
vocabulary, letter-naming, & math
a) Executive functioning (pencil task,
balance beam)
b) Compliance (“cool toys” clean up)
c) Effortful control (giftwrap/no peek +
snack delay)
Plus– Yields ASSESSOR REPORT of
attention/impulse control
Figure 1.3 CSRP Treatment Effects on
Self-regulation and Pre-academic Outcomes
1.0
0.8
Standard Deviation
**
0.6
**
*
0.4
*
*
0.2
0.0
-0.2
Executive
Function
Effortful
Control
Attention/
Impulsivity
-0.4
Raver, Jones, Li-Grining, Zhai, Bub, &
Pressler (2011). Child Development
PPVT
Letter
Naming
Early Math
Skills
** p<0.01, * p<0.05
Are these findings trustworthy?
– Replicate…. Foundations of Learning, Newark & Chicago
• Pamela Morris, PI with MDRC as lead
CSRP
FoL Pilot
Newark
2004-2006
2006-2007
2007-2008
2008-2009
Total # of Sites
18
17
51
20
Program/ Control
9/9
9/8
26/25
10/10
Total # of
Classrooms
36
17
51
40
Timeline
FoL Full-Scale FoL Full-Scale
Newark
Chicago
FOL improved children’s levels of engagement,
behavioral control
Morris, Raver, Millenky, Jones, & Lloyd (2010). Making preschool more productive: How classroom
management training can help teachers. New York, NY: MDRC.
28
Take-home messages of those
intervention findings
• Support for model of experiential canalization of self-reg
(Blair, 2010; Blair & Raver, revise & resubmit).
• When children’s self-regulation is targeted, at the level of
targeted classroom context, children demonstrate
substantial academic gains –
– Vindication of Zigler’s emphasis on “whole child”
• Caution: No “silver bullet” to solve the problems of poverty
– Optimistic hypothesis: children will be better able to capitalize on future
opportunities for learning in early elementary years.
– A less optimistic hypothesis: gains will be sustained only as long as children
continue to have access to high quality classroom practices.
SR and multiple forms of adversity
• RCT great for causal inference, but problem of
limited generalizability (utos)
• Expectations of long-term impact?
• What are children facing, besides ∆ in preschool qual?
Family
poverty &
Instbility
SR and
experimental ∆
in preschool
Family
poverty &
Instbility
SR in K
school
quality
SR in 3rd
school
quality
Children are exposed to “2nd treatment”
of higher vs. lower school quality
SR in the context of K school quality
Treatment
Child Covariates at Baseline
Boy
Age
Non-Hispanic black
Pretreatment score
Low-quality Schools
C-TRF
C-TRF
Internalizing Externalizing
0.91
2.66
(2.26)
(3.39)
High-quality Schools
C-TRF
C-TRF
Internalizing Externalizing
-2.99*
-5.36**
(1.27)
(1.66)
1.71
(2.28)
-0.03
(0.21)
1.45
(1.26)
0.52+
(0.29)
1.34
(2.38)
0.05
(0.27)
-3.50
(3.36)
0.80**
(0.17)
Mother Covariates at Baseline
Mother was married
-2.88
-3.77
(2.64)
(3.37)
Less than high school education
-4.09*
-0.80
(2.01)
(6.41)
Working 10 hours or less per week
3.09+
4.55+
(1.65)
(2.71)
Income below 50% poverty line
-1.35
-0.30
(1.59)
(3.01)
Classroom Covariates at Baseline not included in table for ease of presentation
-1.62
(2.39)
0.10
(0.12)
0.09
(2.08)
0.48+
(0.26)
3.45
(3.20)
0.15
(0.24)
-1.04
(4.89)
0.66**
(0.27)
1.79
(2.68)
1.75
(2.17)
-1.50
(1.61)
1.03
(1.39)
-0.84
(3.33)
1.20
(2.60)
-4.54
(3.49)
-1.08
(3.27)
Overall, the effects
of CSRP on self-reg
and school
readiness in K are
null
Zhai, Raver & Jones (submitted) Children & Youth Services Review using principal score
matching and stratification (n=361) to test for treatment effects in k.
SR and
experimental ∆
in preschool
School
quality K
Where “school quality” = the % of students
meeting or exceeding state standards on the
ISAT within that school.
What about the emotional climate of
the school?
– Children exposed to threatening home
environments have been found to demonstrate
serious disruptions in emotional regulation and EF
– Do these findings extend to CSRP children’s
exposure to unsafe schools?
Capitalizing on Chicago
Student Connection Surveys:
Family
poverty &
Instbility
SR and
experimental ∆
in preschool
Family
poverty &
Instbility
School
quality K
School
quality 3rd
In 2009, 17% of CPS middle
school students reported that
they did not feel physically safe
(problems with fights, theft, or
vandalism)
and/or did not feel emotionally
safe (often teased, bullied,
harassed).
Items included wanting to stay
home, wanting to change
schools due to not feeling safe
at school.
Analytic approach
•
Level-1 equation:
Yij = β0j + β1j(Child’s attention/impulse control in Head Start) + β2j(Child’s gender) +
β3j(Child’s race/ethnicity) + β4j(Child’s grade) + β5j(Chronicity of family poverty) +
β6j(Maternal marital status) + β7j(Maternal education) + β8j(CSRP site pairs) +
β9j(CSRP Tx status) + rij
•
Level-2 equation:
β0j = γ00 + γ01(Unsafe Climate) + γ02(High Adult Support) + γ03(% Low-income
students) + u0j
•
Level-2 equation with attention/impulse control*school climate interactions:
β0j = γ00 + γ01(Unsafe Climate) + γ02(High Adult Support) + γ03(% Low-income
students) + u0j
β1j = γ10 + γ11(Unsafe Climate) + γ12(High Adult Support)
•
Level-2 equation with gender*school climate interactions:
β0j = γ00 + γ01(Unsafe Climate) + γ02(High Adult Support) + γ03(% Low-income
students) + u0j
β2j = γ20 + γ21(Unsafe Climate) + γ22(High Adult Support)
HLM analyses examining child- and school-level
predictors of STRS conflict with teacher in K
Model 1
1.50
(3.03)
Model 2
1.31
(2.95)
Conflict with the teacher in Head Start
0.46***
(0.08)
0.45***
(0.08)
Family income-to-needs ratio at kindergarten
-1.77*
(0.77)
-1.66*
(0.78)
School size (total enrollment)
0.00***
(0.00)
0.00***
(0.00)
% low-income students in the school
-0.11†
(0.06)
-0.08†
(0.05)
Safe and Positive Climate
0.09
(0.78)
1.06
(0.83)
Climate of Respect and Responsibility
-1.27†
(0.72)
-1.82
(1.15)
High Adult Support
2.75
(2.10)
4.44
(3.39)
Safe and Positive Climate*Child is male
Climate of Respect and Responsibility*Child male
-2.56***
1.24
(0.77)
(1.60)
High Adult Support*Child
male
9
-2.78
(4.65)
Conflict with the Teacher
Variable
Child is African American (vs. all else)
8
7
6
5
Girls
4
Boys
3
2
1
0
Low on Safe and Positive Climate
Lowenstein, Raver, Jones, Friedman & Pess (submitted)
High on Safe and Positive Climate
Role of school climate predicting
cognitive regulation, 3rd grade
1.2
Cognitive Dysregulation
1
0.8
0.6
0.4
0.2
Low Att/Imp Cont
0
High Att/Imp Cont
-0.2
-0.4
-0.6
-0.8
Low
Preliminary analyses
suggest:
1) Attention/impulse
control in Head
Start
2) Chronic poverty
(trend)
3) Unsafe School X
Attention/Impulse
control in HS
Med
High
Safe & Positive Climate (Corrected/excludes ISAT)
Expanding perspectives on adversity to include neighborhood
violence
Family
poverty &
Instbility
2500 Feet
2000 Feet
1500 Feet
1000 Feet
1.2
0.8
0.4
Effect size
SR and
experimental ∆
in preschool
0.0
-0.4
-0.8
Neighbrhood
violence
-1.2
-1.6
Executive Functioning
Effortful Control
Attention
Impulse Control
Dependent Variable
Figure 2. Estimates of homicide effects on four dimensions of self-regulation
using homicide zone fixed effects specifications.
Estimates represent effects of homicides occurring within distances of 2,500
feet, 2,000 feet, 1,500 feet, and 1,000 feet of child’s residential address.
Sharkey, Tirado-Strayer, Papachristos, & Raver (to be submitted)
CSRP children exposed to high levels of
violence near home and school
Our current work -- analyses of 5th
grade ER and EF predicted by:
Chronic poverty,
negative school climate, and
household & neighborhood
violence
Family
poverty
Family
poverty
Family
poverty
EF and ER
T
BIS/BRIEF
3rd
EF and ER
5th
RCT
preschool
School
School
Neigh
violence
Neigh
violence
Neigh
violence
School Neighborhood Crime:
Geocoding procedures were
employed to link children’s school
address to the Chicago Police
Department crime index
Implications for prevention
•
Capitalizing on “booster” SEL treatment that occurs nonrandomly in Chicago
Public Schools
–
–
Zhai & Raver, APPAM, 2011
Tradeoffs re: higher external validity, but lower internal validity
•
Capitalizing on out-of-school time: “Adventure Camp” RCT pilot study
– 2010
• N =24, age 7-11, high SES (µ = $99,000)
– 2011
• N =37 , age 7-12, middle to high SES
•
CSRP data collection underway (Cohort 1 collected, Cohort 2 in January!) with
opportunity to pilot, adapt new measures for large-scale survey & prevention
research.
•
How do low-income, ethnic minority 5th graders understand emotion? Deploy
attention when faced with emotionally negative stimuli? Respond to opportunity
to “take gamble?”
•
Testing concurrent and longitudinal relations between ER and EF
Concluding comments
• RCTs as “gold standard” for efficacy of intervention/prevention
• Expanding frameworks to consider the roles of “exogenous
shocks”
– housing vouchers, community policing, income loss
• Capitalize on RCTs for opportunity to draw causal inferences
re: Central role of self-regulation for learning
• Expand basic science models to better understand the role of
adversity in shaping the ER and EF of low-income, ethnic
minority children.
Thanks!