risk and resilience in childhood
TRANSCRIPT
RISK AND RESILIENCE
IN CHILDHOODJo Boyden and Abhijeet Singh
Young Lives, University of Oxford
Rudolph Schaffer:
“Whatever stresses an individual may have encountered in early years, he or she need not forever more be at the mercy of the past. . . . children’s resilience must be acknowledged
every bit as much as their vulnerability”
‘Social Development: an Introduction’ (1996:47)
Understanding risk and resilience
• Risk: a stressor (deficit or ‘insult’) experienced by an individual that
heightens probability of developmental or behavioural pathology
• Cohort studies in LMICs highlight loss of developmental potential due to
risk exposure: • the first 1,000 days of life are critical, when exposure heightens
severity & persistence of effects• synergy between developmental domains compounds effects
• Yet, Rutter (1972) argued that genetically influenced variations in
environmental susceptibility may be important
• ‘Ego resilience’ (Gamezy 1983) :
• good outcomes despite high-risk status • sustained competence under threat • recovery from trauma (Masten, Best, & Garmezy, 1990)
• Rutter has found that risk itself can lead to the development of protective
processes that enhance resilience in children: Steeling effects
Evidence on risk and resilience
• Resilience is highly influenced by “protective processes” in the wider environment (proximal and distal)
• Risks and protective processes are context specific:
• Structural dynamics – risk burden varies widely according to children’s social attributes (ethnicity, religion, gender, caste), location & HH economic status
• Ideational systems – cultural norms & subjective understandings of risk and resilience make a significant difference
• Sociality – relationships and institutional membership are key
• Risks are domain-specific: some competencies (e.g. social & emotional) are far more responsive to environmental stimuli than others (e.g. sensory & motor functions)
• Risks and resilience are both sensitive to timing
• Questions that need answering:
• Is recovery, partial or complete, possible?
• What are the factors behind such recovery? Limitations?
• Which are the critical ages for domain-specific interventions?
• Conversely, which are the most critical domains for each age group?
• Interdisciplinary, mixed methods, comparative study of a dual birth
cohort that is researching the determinants & outcomes of childhood
poverty
• Following nearly 12,000 children in Ethiopia, India (Andhra Pradesh &
Telangana), Peru, Vietnam, started 2001 & ends 2018
• 80 sites selected non-randomly - pro-poor & representing country
diversity (rural-urban, livelihoods, ethnicity etc.)
• Respondents selected randomly as representative of their cohort at
sentinel site level; roughly equal numbers of boys and girls
• 5 survey rounds combine explanatory variables at the child, caregiver &
community level with child outcome indicators ( + measures from
international test batteries) - plus school surveys
• 4 waves of qualitative research with a sub-sample of 200 children
Young Lives
Ethiopia India Peru Vietnam
Younger
cohort (b.
2000/1)
Round 1 (2002)
Round 2 (2006)
Round 3 (2009)
1 year old
5 years old
8 years old
Round 4 (2013) 12 years old
8 years old
12 years old
15 years old
19 years old
Round 5 (2016) 15 years old 22 years old
Older
cohort (b.
1994/5)
Full sample of
children &
caregivers, plus
selected younger
siblings &
community
representatives
Structure of panel
Example 1: Catch-up from stunting
• Children in developing countries born close to developed
country norms but falter in the first two years of life
(Victora et al. 2010)
• Faltering is often thought to be irreversible
• Panel data show however that a number of stunted
children do ‘catch-up’ i.e. recover from stunting.
• Cebu (Adair, 1999), Young Lives in Ethiopia, India, Peru and
Vietnam (e.g. Crookston et. al. 2010, Crookston et. al. 2013)
• Surprisingly children who have recovered from stunting
and never stunted show no difference in cognitive skills
(Crookston et. al. 2010)
• So, perhaps stunting not as hopeless as believed.
Catch-up growth: Why it matters
• Factors that correlate with catch up are similar to those which prevent stunting in the first place• The initially worst-off are the most vulnerable (similar to risk in
other domains, e.g. Dercon 2002)
• Those stunted earliest least likely to catch-up (Adair, 1999)
• Policy may have a role: • early remediation most effective, but some hope still for those
left behind
• A concrete illustration: Singh, Park and Dercon (2014) show that India’s Midday Meals help compensate for nutritional effects of droughts in early childhood• So catch-up may be manipulable by policy tools.
• Focusing entirely on risks denies possibility of remediation
Understanding the ages for intervention:
learning divergence• Two key questions relevant for child outcomes:
• When do gaps emerge between advantaged and
disadvantaged children?
• This may indicate where remediation/prevention should focus
• What are the sources of the gaps?
• Indicates which domains should be targeted.
• I study these questions focusing on learning divergence
between children in four YL countries
• Learning levels differ widely across countries and with
direct relevance to growth and inequality
Learning from 5-8 years
p10 p90
400
450
500
550
60
0M
ath
score
s (
20
09
)
300 400 500 600 700CDA scores (2006)
Ethiopia India
Peru Vietnam
Timing and sources of gaps
• 5-8 years a period of major divergence in learning levels• For observationally equivalent children
• The quality of schooling, measured as learning gain per year of schooling, the major source of divergence:• Equalizing that wipes out divergence between Peru and Vietnam
• Closes 70% of divergence between Vietnam and India
• (Results based on value-added models and regression discontinuities)
• Why this is important:• Early years undoubtedly important (achievement at 5 an important
factor) but not all lost
• Large randomized experimental literature explores channels of intervention
• Panel-based overview highlights possibilities and domains of interventions – a conceptual underpinning of risk and resilience.
Understanding gender gaps: Dercon and Singh
2013
• Gender differences frequently context-specific:• Academic outcomes pro-boy biased in India and Ethiopia, pro-girl
biased in Vietnam, and no distinguishable bias in Peru
• Frequently domain-specific: • Nutritional outcomes (esp. BMI-for-age at15) pro-girl biased in most
countries including where educational investments are distinctly pro-boy biased
• Changing in nature:• A decade ago, gender bias in India was on enrolment (Kingdon 2005)
but now it is in private schooling (Maitra et. al.)
• Dynamic in emergence:• Agency at 12 predicts agency at 15 and educational outcomes
• The type of foundational work that observational data (esp. panels) yield which precede (experimental) program-based planning
Coming back to the questions
• Is recovery, partial or complete, possible?
• What are the factors behind such recovery?
Limitations?
• Which are the critical ages for domain-specific
interventions?
• Conversely, which are the most critical domains
for each age group?
• These should be questions where there is a
clear advantage for panel-based designs.