Presentation on theme: "Total Family Impact on Status Attainment - Sources of Sibling (Dis)similarity Social Stratification Research Conference, Utrecht September 10, 2010 Antonie."— Presentation transcript:
Total Family Impact on Status Attainment - Sources of Sibling (Dis)similarity Social Stratification Research Conference, Utrecht September 10, 2010 Antonie Knigge, Ineke Maas, Marco van Leeuwen
Background Towards Open Societies? Occupational status father as indicator for family impact Underestimates the true influence of the family Genes, socialization, financial resources, social capital, etc. Problem: how to measure total family impact? Impossible to measure all aspects Traditional measures explain about 60% of total family impact Solution: sibling models the more similar siblings in status compared to unrelated persons, the larger total impact of the family background Intra-class correlation (ICC) as measure
Background (2) Aim: Examine whether sibling models are a valid tool for assessing (trends in) the total family impact on status attainment Implicit Assumption: siblings benefit equally from the resources of their parents Hunch: not always realistic Example: equal vs unequal inheritance
Inheritance practices in 19th/early 20th century the Netherlands Partible & Equal Impartible & Equal Impartible & Unequal
Research Question Does the extent to which siblings benefit differently from the resources of their parents form part of the explanation of trends in siblings status similarity for different regions in the Netherlands from 1842 to 1922?
Theory We formulate conditions under which we expect siblings to benefit systematically different from the resources of their parents Example Hypothesis: H5. Siblings with parents who are land-owning farmers will be less similar in their attained status in communities with an unequal inheritance system than in communities with an equal inheritance system
Data: Genlias Information from around 600.000 Dutch marriage acts covering 5 of 11 provinces during the 1842-1922 period Only look at grooms Information on Son + Father occupation, place & year of birth, place & year of marriage Marriage act groom linked to marriage act parents We know the married siblings of a groom Complemented with other historical sources Dutch Bur. Statistics, archives, etc.
Strategy: Multilevel sibling models Standard multilevel sibling model: siblings nested in families Grooms also share a context Add cross-classified levels for time and place Auto-correlative structure for time and place Families F1 F2 F3 Siblings Families F1 F2 F3 P1 P2 T2T3P3 T1 Plaats & Tijd Siblings
Standard model Dependent variable: Occupational Status Groom Constant47.11 Independent Variables Occ status father.48 Sibsize-.10 Etc. Variance Components Sibling level45 Family level65
Standard Model + Extension 1 Dependent variable: Occupational Status Groom Constant47.11 Independent Variables Occ status father.48 Sibsize-.10 Etc. Variance Components Sibling level45 Family level40 Place level20 Time Level5
Strategy: Extension 2 Standard multilevel sibling models: variance components of each level are homogeneous According to hypotheses, we expect them to be heterogeneous Siblings less similar in unequal than siblings in equal inheritance farming families Explicitly model the variance components at sibling and family level
Standard model Dependent variable Occupational Status Groom Constant… Independent Variables… Variance Components Sibling level45 Family level65
Standard model + Extension 2 Dependent variable Occupational Status Groom Constant… Independent Variables… Variance Components Sibling level40 Equal inheritance400 Unequal inheritance5010 Family level63 Equal inheritance630 Unequal inheritance674
Results Not succeeded in both extensions at the same time First: Extension 1 Second: Extension 2 Compare results both Results not so important More important: what to compare?
Cross-classified Multilevel Models for farming and non- farming background and different inheritance practices Dependent: Occ. status groom Father Non-farmerFather Farmer Partible, Equal Impart., Equal Impart., Unequal Partible, Equal Impart., Equal Impart., Unequal Constant53.5450.0146.6153.6451.0347.68 Var(sibling)71.9767.3869.6166.4853.5755.33 Var(family)93.3774.5265.8819.9816.4210.74 Var(cohort)5.932.5220.127.116.111.02 Var(region)84.0657.710.7868.9642.622.50 ICC.18.104.22.168.16 N(siblings)311741071015965790673560842132 N(families)18358625013552851721964324307 N(cohorts)18 N(regions)2711557218414671
2-level Multilevel Model: heteroskedastic variances for farming background and inheritance practice Dependent: Occ. Status groom Constant47.48 Variance SiblingFamily Constant71.3388.18 Partible equal (non-farming)15.6844.15 Impart. equal (non-farming)8.2426.07 Impart. Unequal (non-farming)00 Farming, parible equal13.75-84.09 Farming, impartible equal-15.39-89.86 Farming, impartible unequal-15.00-74.52 N564686324382
Conclusion & Discussion It seems promising to model variance components to be heterogeneous Sources of heteroskedasticity not always clear Results sometimes not in line with hypotheses Issues to explore: Deepen out historical context Disentangle openness from other sources of (dis)similarity What is the right measure: ICC or something else? Non-random missings