Death miracles and resurrection Fadeout in human capital interventions Death miracles and resurrection Impact Death IQ impacts in Perry Achievement impacts for Head Start 3 year olds Cognitive impacts in 67 ECE studies ID: 745258
Download Presentation The PPT/PDF document "Fadeout in human capital interventions:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Fadeout in human capital interventions: Death, miracles and resurrection
Fadeout in human capital interventions: Death, miracles and resurrectionSlide2
Impact
DeathSlide3
IQ impacts in PerrySlide4
Achievement impacts for Head Start 3 year oldsSlide5
Cognitive impacts in 67 ECE studiesSlide6
Employment
impacts for the 36-month Canadian Self-Sufficiency ProgramSlide7
Impact
PersistenceSlide8
IQ impacts in Perry and AbecedarianSlide9
IQ and Earnings impacts in Perry
A “noncognitive” miracle occurs…Slide10
Earnings impacts for the Job Training Partnership Act ProgramSlide11
Patterns of fade-out
A mess!
Impacts fade out in some interventions but don’t in seemingly similar interventions
Sometimes, for the same intervention, some impacts fade out but others emerge decades laterSlide12
Outline
OPTIMAL CONDITIONS FOR IMPACT PERSISTENCE
OTHER AVENUES FOR IMPACT PERSISTENCE
HOW TO RECONCILE ECE PROGRAM FADEOUT ON IQ WITH IMPACTS ON ADULT OUTCOMES?Slide13
Outline
OPTIMAL CONDITIONS FOR IMPACT PERSISTENCE
OTHER AVENUES FOR IMPACT PERSISTENCE
HOW TO RECONCILE ECE PROGRAM FADEOUT ON IQ WITH IMPACTS ON ADULT OUTCOMES?Slide14
What conditions lead to impact persistence?
the “right kinds” of skills or capacities, or
the “right kinds” of environments
When interventions change:Slide15
What conditions lead to impact persistence?
the “right kinds” of skills or capacities, or
the “right kinds” of environments
When interventions change:Slide16
“Right kinds” of skills:
Skills or behaviors
fundamental for success
in adulthood or for childhood attainments…
that are
malleable
…
and
would not develop eventually
in counterfactual conditionsSlide17
“Right kinds” of environments:
Malleable features of environments that are
fundamental for
promoting
the “right kinds” of skills and behaviorsSlide18
Fundamental and malleable skills
Fundamental
Peripheral
More malleable
Less
malleable
Conscientiousness (grit)
g
(IQ
)
Who cares?
Teaching to the test
SAT
test
prep
Flash cards
FAFSA rule knowledge
?...Slide19
Fundamental AND malleable skills?
Math
Literacy
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
Executive function
Emotional self-regulation
Background knowledge
Fixed
vs. malleable
intelligence (Dweck)
Self-worth (Cohen
et al.)
Math: number line, fractions, algebra
Literacy
Background knowledge
Executive function
Prosocial behaviorsSlide20
“Right kinds” of skills:
Skills or behaviors
fundamental
for success in adulthood or for childhood attainments…
that are
malleable
…
and would not develop eventually in counterfactual conditionsSlide21
Fundamental AND malleable skills?
Math
Literacy
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
Executive function
Emotional self-regulation
Background knowledge
Fixed
vs. malleable
intelligence (Dweck)
Self-worth (Cohen
et al.)
Math: number line, fractions, algebra
Literacy
Background knowledge
Executive function
Prosocial behaviors
Which would develop eventually and therefore generate impact fadeout?Slide22
Pace of development in counterfactual conditions
Null/slow
=> no fadeout?
Eventually
=> fadeout?
Math
Literacy
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
Executive function
Emotional self-regulation
Background knowledgeSlide23
Pace of development in counterfactual conditions
Null/slow
=> no fadeout?
Eventually
=> fadeout?
Fractions, algebra
Large vocabulary
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
EF working memory
Emotional
self-
reg
for some
Background knowledge
Counting
Alphabet knowledge
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
EF impulse control
Emotional
self-
reg
for most
Background knowledgeSlide24
Examples of impact persistence
Algebra
Chicago’s double-dose algebra
Self-concept
Value affirmationSlide25
Persistent impacts of 9th grade double-dose algebra in Chicago
Source: Cortres et al. (2011)
Impact
B or higher in 9
th
-grade algebra
+13% *
A in 9
th
-grade algebra
ns
Passed geometry
in 10
th
grade
+12%
*
Grade 11 math scores
+.24
sd
*
Graduated
within 5 years
+12
% *
Enrolled in any college
+11%
*Slide26
Cohen values affirmation impacts on low-GPA Black students
Program periodSlide27
Cohen caveats
No impacts on higher-achieving Blacks and whites
Some replication attempts show no consistent impacts (Dee, 2014)Slide28
Conditions where “eventual”
development in counterfactual may lead to fadeout
Impulse control by age
Did the Canadian SSP accelerate return to labor force that would have happened anyway?Slide29
Distraction time by age
Posner and
Rothbart
(2007)Slide30
Did Canadian SSP speed up employment that would have occurred anyway?Slide31
Did Canadian SSP speed up employment that would have occurred anyway?Slide32
Did Canadian SSP speed up employment that would have occurred anyway?Slide33
What conditions lead to impact persistence?
the “right kinds” of skills or capacities, or
the “right kinds” of environments
When interventions change:Slide34
“Right kinds” of environments:
Malleable features of environments that are
fundamental for
promoting
the “right kinds” of skills and behaviorsSlide35
School quality can be a fundamental environmental feature
Winning the lottery to enter one of NYC’s small high schools of choice
(
Untermann
et al., 2014)Slide36
Outcomes
SSC
Control group
Effect
Graduation
Graduated from high school
71.6
62.2
9.4
**
Regents diploma granted
50.2
43.5
6
.7
**
Advanced Regents diploma
granted
8.2
7.3
0.9
College readiness
Passed English
Regents Exam at 75+
42.
1
35.8
6.3
**
Passed Math
Regents Exam at 75+
25.1
24.5
0.5
Post-secondary enrollment
Enrolled in post-secondary education
49.0
40.
7
8.4
**
Winning the lottery to enter a NYC Small High School of Choice (n=14,608)Slide37
Is neighborhood quality a fundamental environmental feature?
Moving to Opportunity suggests not for many outcomes in the USSlide38
MTO: Huge Differences in Neighborhood Poverty
(Duration-Weighted)
38Slide39
Outcome
Interventions Impacts
Baseline Ages 0 to 5
Reading Assessment
ns
Math Assessment
ns
Baseline Ages 6 to 11
Reading Assessment
ns
Math Assessment
ns
Took SAT/ACT?
ns
No Impacts
on
School Achievement
39Slide40
What else can sustain impacts?
When interventions:
(from before)
boost the “right kinds” of skills or
environment
are supported by post-TX sustaining environment
lead
to foot-in-the-door access to sustaining environments
are sufficiently intensive to change foundational skills for children with bad counterfactual conditions
treat enough children to generate positive peer effectsSlide41
What can sustain impacts?
When interventions:
(from before)
boost the “right kinds” of skills or
environment
are supported by post-TX sustaining environment
lead
to foot-in-the-door access to sustaining environments
are sufficiently intensive to change foundational skills for children with bad counterfactual conditions
treat enough children to generate positive peer effectsSlide42
Building Blocks and sustaining environmentsSlide43
Building Blocks and sustaining environments
ns
Similar results for low vs. high math home environmentsSlide44
Building Blocks and sustaining environments
ns
p
=.07Slide45
Sustain environments
Building Blocks suggest that environmental supports must be tailored explicitly to the nature of the prior treatmentSlide46
What can sustain impacts?
When interventions:
(from before)
boost the “right kinds” of skills or
environment
are supported by post-TX sustaining environment
lead
to foot-in-the-door access to sustaining environments
are sufficiently intensive to change foundational skills for children with bad counterfactual conditions
treat enough children to generate positive peer effectsSlide47
Foot-in-the-door examples?
SAT prep may affect college quality, which is known to have a positive impact on
earnings
Can FAFSA knowledge lead to college and later success?
Was some of Abecedarian’s long-run success caused by lower rates of special
ed
and grade retention?
Can pro-social behavioral interventions reduce or delay first arrests?Slide48
Foot-in-the-door
Foot-in-the-door
links
to the emerging literature on
developmental cascades
(Dodge et al. 2008)
Sequence of positive or negative conditions that cumulate to good or bad outcomes
Since <100% probabilities multiply, relying on cascades seems like a risky intervention strategySlide49
What can sustain impacts?
When interventions:
(from before)
boost the “right kinds” of skills or
environment
are supported by post-TX sustaining environment
lead
to foot-in-the-door access to sustaining environments
are sufficiently intensive to change foundational skills for children with
difficult counterfactual
conditions
treat enough children to generate positive peer effectsSlide50
IQ impacts in Perry and AbecedarianSlide51
Perry vs. Abecedarian
1 or 2 years
Part-day ECE + weekly home visits
Urban setting
5 years
Year-round full-day ECE
Cumulative curriculum
Rural setting
Perry
Abecedarian
African
American children with low tested
IQs
High risk/low SES familiesSlide52
What can sustain impacts?
When interventions:
(from before)
boost the “right kinds” of skills or
environment
are supported by post-TX sustaining environment
lead
to foot-in-the-door access to sustaining environments
are sufficiently intensive to change foundational skills for children with bad counterfactual conditions
treat enough children to generate positive peer effectsSlide53
Generating peer effects
Measles vaccination!
Deworming treatments in Kenya generated large benefits for untreated children in treated schools (Miguel and Kremer, 2004)
County spending on preschool?Slide54
More county spending on ECE boosts grades 3-5 achievement
Ladd, Dodge, &
Muschkin
(2014, JPAM),
Muschkin
, Dodge, & Ladd (in press, EEPA), Dodge, Ladd,
Muschkin
, & Bai (under review)Slide55
Outline
OPTIMAL CONDITIONS FOR IMPACT PERSISTENCE
OTHER AVENUES FOR IMPACT PERSISTENCE
HOW TO RECONCILE ECE PROGRAM FADEOUT ON IQ WITH IMPACTS ON ADULT OUTCOMES?Slide56
Death
ResurrectionSlide57
A miracle occurs…Slide58
Perry and Abecedarian both affected adult earnings
Do they have common (statistical) mediators?
Are they “cognitive” or “non-cognitive”?
ECE treatment
Mediators
Adult earningsSlide59
Perry’s IQ Effects by Age
Source:
Schweinhart
et al., 2005; Effect sizes >.30 are p<.05, one-tailed testSlide60
Perry’s IQ and Achievement Effects by Age
Source:
Schweinhart
et al., 2005; Effect sizes >.30 are p<.05, one-tailed testSlide61
Perry’s Noncognitive Effects Ages 6-9
Source: Pinto, based on Heckman et al. (2014)
Index
Sample item
Impact
Academic potential
Creativity
.31
ns
Academic motivation
Alert
and interested; motivated; persists
.37
ns
Classroom conduct
Disobedient;
impulsive; blames others
.40*
Personal behavior
Absences;
swears
.36
ns
Teacher dependence
Seeks constant reassurance
.03
ns
Emotional state
Depressed;
withdrawn
.29
ns
Emotional
adjustment
Trust; level of emotional adjustment
.30
nsSlide62
Perry (statistical) mediators
Perry generated a raft of potential mediational effects
achievement, but not IQ
a number of potentially important “
noncognitive
” domainsSlide63
Abecedarian’s IQ and Achievement Effects by Age
Source: Campbell et al., 2001; all effect sizes are p<.05.Slide64
Abecedarian’s IQ and Achievement Effects by Age
Source: Campbell et al., 2001; all effect sizes are p<.05.Slide65
Abecedarian’s Cognitive and Noncognitive Effects by Age
Source: Campbell
et al. (2001) and Campbell et al. (2002)Slide66
Abecedarian (statistical) mediators
Abecedarian generated a number of potential mediational effects
achievement and IQ
but not a limited number of “
noncognitive
” measuresSlide67
Two other relevant studies
Both the Chicago Parent-Child program (Reynolds et al.) and kindergarten class quality (
Chetty
) affected adult
earnings
CPC affected reading and math achievement more than its “
noncognitive
” measures
Chetty
et al. found more impact on a
noncognitive
index than on achievementSlide68
Bottom lines on impact resurrection
An even bigger mess!
Q: haven’t developmental psychologists invented a word for a big mess like this?
A: YES – “
equifinality
,” when many roads lead to the same outcomesSlide69
Summary
Impact persistence requires treating
fundamental
and
malleable
skills that
would not develop eventually
in counterfactual conditions
Other avenues are possible
Foot-in-the-door cascades from peripheral skills are possible but risky
Environmental or intensive individual interventions may work but are expensive
Targeting children in the worst counterfactual conditions may be the best strategySlide70
Bottom lines
Research priorities
Long-run follow-ups, perhaps using administrative data
Design better
post-TX sustaining environment
For interventions involving implicit theories, self-concept and motivation, we need more independent replications and longer-run follow-upsSlide71
That’s it!Slide72
Fundamental and malleable skills
Fundamental
Peripheral
More malleable
Less
malleable
Conscientiousness (grit)
g
(IQ
)
Who cares?
Teaching to the test
SAT
test
prep
Flash cards
FAFSA rule knowledge
?...Slide73
Pace of development in counterfactual conditions
Null/slow
=> no fadeout?
Eventually
=> fadeout?
Fractions, algebra
(much) Vocabulary
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
EF working memory
Emotional
self-
reg
for some
Background knowledge
Counting
Alphabet knowledge
Implicit theories (Dweck)
Self-concept (Cohen)
Academic motivation
EF impulse control
Emotional
self-
reg
for most
Background knowledge