Presentation on theme: "HEREDITY, FAMILY & INEQUALITY Michael Beenstock Hebrew University of Jerusalem January 2011."— Presentation transcript:
HEREDITY, FAMILY & INEQUALITY Michael Beenstock Hebrew University of Jerusalem January 2011
Francis Galton Died 1911 Discovered regression and correlation 1884 Regression towards the mean in height Beta convergence and sigma convergence Founder of behavioral genetics: Twin studies Founder of eugenics, fingerprinting and IQ tests Nature v Nurture: Pro nature & heredity
Correlation within the Family Intergenerational: parents & children Intragenerational: siblings Anthropometry: birth weight, height, BMI Demography: fertility, longevity, IQ Social: crime, religiosity, personality Economics: schooling, earnings, wealth Earnings: consensus estimate 0.4.
Inequality beta & sigma convergence Y c = + Y p + c 0 < < 1 Var(Y c ) = 2 var(Y p ) + var( c ) Var(Y) →var( )/(1 - 2 )
Nature v Nurture Aristotle & Plato: Nature + Heredity Locke: tabula rasa (blank slate) v innatism Darwin: back to nature Developmental psychology: nurture Behavioral genetics: nature Economics: mainly nature
Innatist Controversies Jenkins (1968) Race & IQ Wilson (1975) Sociobiology & Humans Dawkins (1976) Selfish Gene Herrnstein & Murray (1994) Bell Curve: IQ matters more than family Harris (1998) Nurture Illusion
Behavioral Genetics P = G + E Var(P) = var(G) + var(E) + 2cov(GE) r GE =cov(GE)/sd(G)sd(E) gene-environment correlation h 2 = var(G)/var(P) heritability (nature) e 2 = var(E)/var(P) (nurturability) Cov(P 1 P 2 )=cov(G 1 G 2 )+cov(E 1 E 2 )+2cov(E 1 G 2 ) r = h 2 r G + (1 – h 2 -2her GE )r E + 2her G1E2
MZ & DZ Twins r G.MZ = 1 r G. DZ = ½ r E.MZ = r E.DZ Equal Environments (EEA) r GE = 0 r G1E2 = 0 h 2 = 2(r MZ – r DZ ) r MZ = 0.7 r DZ = 0.3 h 2 = 0.8 Alternative assumptions: h 2 = 0! No behavioral theory in behavioral genetics
Adopted & Biological Siblings r G.A = 0 r G.B = ½ r E.A = r E.B (EEA) r GE = 0 etc h 2 = 2(r B – r A ) Random selection into adoption!
Methodological Problems Genotypes not observed: omitted variable bias Environments endogenous: simultaneous equations bias Environments depend on unobserved genotypes: E(Xu) ≠ 0 I am my sibling’s sibling: Reflection Bias Solution: IV for environments
Natural Experiments Environmental effects Kling et al (2005) Moving to Opportunity Oreopoulos (2003) Toronto Edin at al (2003) Immigrants in Sweden Gould et al (2004) Immigrants in Israel Shea (1999) Parents’ income in US Beenstock (2010) Parents’ schooling and income in Israel Small effect sizes
Measuring Genotypes Main problem: G is not observed Y = + E + θX + u u = λG + e E(Eu) = λE(EG) Generated Regressor Methodology: Beenstock (2007, 2008, 2010) Mincer residuals measure unobserved earning ability
Genome-wide Association Studies (GWAS) Human Genome Project 2003: Mapping DNA 23 chromosomes 3 billion base pairs Quantitative trait loci (QTL) unobserved Use genetic markers (SNPs) to measure QTLs 1 million markers N = 20,000 Data-mining: low statistical power Bonferroni t – statistics reduce false positives Medicine: replication failure Induction (Bacon) v deduction (Hume)
GWAS & Social Science (Geneonomics)? Longitudinal Study of Adolescent Health Health Retirement Survey Wisconsin Longitudinal Study A gene for violence? Caspi et al (2002) A gene for schooling? Beauchamp et al (2011) A new age is not dawning
Nature – Nurture: So What? Blank slaters insist the slate is 100% blank Innatists don’t insist slate is 0% blank If slate is 0% blank: genetic determinism Social intervention can’t help Research so far suggests that environments don’t matter much Does not mean that genes matter by default