1. Large databases vs. individual analysis: Two complimentary approaches in the study of education and learning Esther Adi-Japha School of Education, Bar-Ilan University, Ramat- Gan, Israel.

2. Studying Effects of Early Education and Care on Child Developmental Outcomes Experimental studies – random assignment; model programs Quasi-experimental studies – treatment & comparison groups; large-scale publicly funded interventions - Experimental & Quasi experimental – low income, at risk families Correlational studies – naturally occurring variations

3. Experimental studies

4. Quasi-experimental studies

5. Correlational studies

6. The NICHD Study of Early Child Care and Youth Development Large sample (n = 1364) from 10 sites Quality, amount, and type of child care measured from birth to kindergarten Mothers and fathers observed and interviewed Home observations Cognitive, language, and social development assessed Children studied from birth to age 18 years (began 1991)

7. Political/Policy issues What can children benefit from high quality child care? Is child care a risk factor for diminished cognitive and social child outcomes? Studying Effects of Early Education and Care on Child Developmental Outcomes

8. Fundamentals of high-quality care A high adult-child ratio Small size group Post-secondary training/education Positive caregiver-child relationship Well defined spaces Well structured, well planned curricula

9. Age Staff:child ratioMaximum group size In Canada (Province of Ontario) Under 18 months1:310 18-23 months 1:5 15 2-5 years1:8 16 In the United States, states may set ratios and maximum group sizes, although many do not set standards for group size. Pennsylvania, for example Rec 6 and 15 months1:3-4 6-8 24 months1:4-58-10 3 years1:6-718-21 14 5 years1:824 16 7 years1:1224

10. On the NICHD SECCYD higher child care quality predicted … Higher cognitive skills at 15, 24, 36, and 54 months and in first grade Higher academic skills at 36 and 54 months Higher language skills at 36 & 54 months Reduced behavior problems Effects of child care quality were larger for children of low-income families. (derived EHS)

11. -Controlling for site, child ethnicity, child gender, maternal education, mean income-to-needs ratio between 6 months and assessment age, parenting quality between 6 months and assessment age, partner status. -Implementing complicated imputation scheme (MI)

12. Conclusion Positive effects for high quality child care But what about the effect of other aspects of child care: e.g., quantity and type of child care?

13. Quantity of care Repeated separation from mother –Might affect infant and toddler attachment to mother, social and cognitive development Long repeated separations, –Might increase stress and lead to behavior problems (unfriendly assertiveness, noncompliance and aggression)

14. 36 to 54 month olds18 to 35 month olds Hours of Care 1 to 17 month olds Hours per week 0 to 10 >10 to <30 30 to highest

15. Type of care Highly structured, school-like rather than home-like environment –Might lack in emotional support Unstructured, home-like environment –Might lack in cognitive and social stimulation

16. * Exclusive maternal care did not predict child outcomes, * Higher quality child care was related to advanced cognitive, language, and preacademic outcomes at every age and better socioemotional and peer outcomes at some ages. * More childcare hours predicted more behavior problems and conflict, according to care providers. * More center-care time was related to higher cognitive and language scores and more problem and fewer prosocial behaviors, according to care providers. NICHD SECCYD Results of 15 - 54 months

17. -Controlling for site, child ethnicity, child gender, maternal education, mean income-to-needs ratio between 6 months and assessment age, parenting quality between 6 and assessment age, partner status. -Implementing complicated imputation scheme (MI) NICHD SECCYD, 2006, American Psychologist

18. Long term durable effects: 6 th grade * Higher quality care predicted higher vocabulary scores * More exposure to center care predicted more teacher- reported externalizing problems. Controlling for preschool time invariant covariates included site, child ethnicity, child gender, maternal education, mean income-to-needs ratio between 6 and 54 months, parenting intercept, and slope from 6 to 54 months, maternal depressive symptoms intercept and slope from 6 to 54 months. The concurrent time-varying covariates from 54 months through sixth grade included income to-needs ratio, parenting, maternal depression observed school classroom quality, and hours per week of after-school care (set to 0 for 54 months). Belsky et al., & NICHD SECCYD, 2007, child development.

19. What are child care effect sizes - Overall, modest to moderate Child care quality Modest effects on all cognitive outcomes (.08 < r p <.12) Modest to moderate on many social–emotional outcomes (.08 < r p <.12) Long term effects on academic scores (6 th grade) Quantity of care Negligible effect on cognitive outcomes Modest negative effects on many social–emotional outcomes (.09 < r p <.14) No long term effects. Center care hours Modest effects on all cognitive outcomes Modest negative effects on many social–emotional outcomes Negative long term effects on social-emotional measures. Parenting Moderate-to-large associations with all cognitive outcomes (.17 < r p <.34;.40 < d < 1.23), many of social–emotional outcomes (.08 < r p <.23;.33 < d <.83) Parenting effects are a twice to three times larger than the corresponding child care effects

20. Are child care effects independent of parenting quality? In contrast to previous (less controlled) studies, the NICHD SECCYD did not find support for 1. ‘The differential prediction hypothesis’. 2. ‘The compensation/lost-resources hypothesis’. But what about non-linear associations ? The role of secondary analysis!

21. Research Question Is there a difference between ‘high-quality’ and ‘low-quality’ parenting outcomes that depends on quantity of care? Is such a difference non-linear?

22. Study Variables Quantity of care - Predominantly maternal care 0-10 hr/week on average 0-36 months - Medium amount of child care 10-32 hr/week on average 0-36 months - Higher amounts of child care Parenting quality ‘low-quality’ and ‘high-quality’ parenting: above and below median Outcomes: Cognitive outcomes at 36 months Prof. Pnina Klien

24. -Controlling for child ethnicity, child gender, maternal education, mean income-to-needs ratio between 6 months and assessment age, parenting quality between 6 and assessment age, partner status. -Implementing complicated imputation scheme (MI) - Results remained with different cut-offs, and with childcare hours as a continuous variable.

25. Conclusions There is an optimum of child care hours, in which the association between parenting quality and outcomes is maximal; associations do not decrease linearly with the amount of child care. Sharing databases is highly important for studying in-depth the results, highlighting un- expected effects.

26. C. To avoid inadvertent disclosure of persons, families, households, or care providers information … … 2. In no case should the total figure for a row or column of a cross- tabulation be fewer than three. 3. In no case should a quantity figure be based upon fewer than three cases. 4. In no case should a quantity figure be published if one case contributes more than 60 percent of the amount. 5. In no case should data on an identifiable case, nor any of the kinds of data listed in preceding items 1-3, be derivable through subtraction or other calculation from the combination of tables released. Can individual differences be studied using large databases? - well, probably not… From the NICHD SECCYD data contract

27. Why should we study individuals’ data? -Group data may mask abrupt changes in the individual’s course of development. These changes may/may-not-be at different times for different individuals. I shall Illustrate this point using a procedural learning task

28. What is procedural learning? Procedural learning refers to the acquisition of new behaviors through a process of repetitive practice (e.g., writing, driving, playing a musical instrument etc..). Laboratory- motor learning tasks are often used as a model for procedural learning.

30. Motor learning Brashes- Krug, Shadmehr & Bizzi, Nature, 1996 Karni et al., PNAS, 1998 Sosnik et al., Exp. Brain Res. 2004

31. Three stages in motor learning Fast learning – gains (in speed and accuracy) accrued throughout the training session. Memory consolidation – gains that appear in the hours or days after termination of training. Retention/Slow learning – retention/further improvement that appear some weeks following termination of practice.

32. 0 0.2 0.4 0.6 0.8 # errors 24 h48 h6 wks training session 9 year-olds 12 year-olds 17 year-olds Performance (# seq./30sec) 3 5 7 9 11 13 15 17 19 21 23 1 Three stages in motor learning

33. Interference Brashes- Krug, Shadmehr & Bizzi, Nature, 1996

34. 0 5 10 15 20 25 0 0.5 1 init end 24hr post init end 24hr post 9 yrs 12 yrs17 yrs no interference interference # errors Performance (# seq./30sec) init end 24hr post Dorfberger, Adi-Japha, & Karni, 2007, Plos one

35. Consolidation gains - 24 hours after termination of training - sleep dependent. Changes may/may-not-be at different times for different individuals…

36. What happened within the training session?

37. Var(moving window(Data - Power-law). Three phases: (a) Low var. (b) High var. (c) Low var. Performance > Power-law extrapolation(1 st Low var.) Adi-Japha et al., 2008, JEP LMC

38. Two performance phases within the training session - Within/between sequence errors - Latency to respond

39. Final conclusions -For the benefit of all, (large) databases need to be shared -Large educational databases are limited: do not allow the study individual behaviors