School Dropout in Rural Vietnam: Does Gender Matter? Jianye Liu Population Study Center Sociology Department of UWO Why this topic?     Education is not only a human right, but is also crucial to national, familial, and individual development It is a central concern in every developing country While the enrollment of girls has increased in past decades, girls continue to be less likely to complete their schooling Why Vietnam? Cultural practices • The Chinese Occupation of Vietnam: -Confucianism: women have lower status than men both in family and society -Patriarchal, Patrilineal and Patrilocal society • The French Occupation and then socialism cancelled out many of the values of Confucianism • Socialist reforms (market-oriented) since 1986 resumed many of the values of Confucianism - the kinship system - son preference Economic situation: Vietnam is a poor country: a typical pyramid structure of education system (no enough space for children in schoool) Recent economic reforms: school system is mainly funded by local governments and the family The household average expenditure on schooling has more than doubled in the 1990s The family is the basic productive unit: More child labor is needed both inside household and in the farm work Results in a relatively high “school dropout rate” Theoretical framework Data and Methodology Data           Vietnam Living Standard Surveys (VLSS) conducted in 1997-98, a state representative survey Two sets of questionnaires: Individual Commune and school (minor urban and rural areas) Stratified cluster sample: Strata: Urban/rural; Cluster: community; Household (all individuals) Children aged to 18 ever in school are used in the analysis Methodology: Multilevel Model Why?  Theoretical reasons: effects of social context: the community and school characteristics affect individual behavior Cross-level interactions are theoretically necessary Unobserved heterogeneity: impossible to include all theoretically related variables into the model (individual motivation for school) Statistical reasons:      Cluster samples (as with the VLSS) violate: Methodology: Multilevel Model  Dependence: Individuals in the same cluster will be correlated because they experience common social settings  Heteroscedasticity: variances are different in different clusters  The regression coefficients produced by OLS: biased because of heteroscedasticity standard errors will be too small because of dependence ( Type І error)    Without taking into account this clustering of cases in my sample, the inferences will be incorrect Operation of Multilevel Model   Program: HLM (Hierachical Linear Model) Two level model:    level-1: individual (6816 cases) level-2: school (532) level-2: community (158 cases)  Dependent variable:  whether a child ever in school dropped out of school   0= in school or have a secondary school diploma; 1= out of school without a secondary school diploma Results Empty (unconditional) model   To gauge the magnitude of variation between communities in school dropout  Conclusion:  School dropping out rate varied from community to community social and school contexts have significant effects Multilevel model has to be used for the data analysis   Conditional models  Individual level:   Gender matters: boys are less likely to drop out of school (almost half of the probability of girls), with all other things equal The probability of school dropout increases with age and children’s labor participation •Household level The odds of school dropout decrease with the household income and the proportion of household expenditures on education The probability increases with the number of siblings in the household, but decreases with the household size The household head’s education level is positively related to children staying in school Commune/ward context: Whether the commune/ward has any traditional occupations or handicrafts increases the occurrence of school dropout The probability of school dropout varied with economic level of commune/ward School context: The pyramid structure of school system results in a strict selective process in children’s schooling The probability of school dropout increases with the school fees Interaction term: The interaction between gender and socioeconomic level of commune indicates that a gender difference in school dropout is more likely to appear in the undeveloped areas, narrows with the increase of human development level, and disappears in the highly developed areas The significant interaction between gender and the quality of teachers indicates that the gender gap in school dropout changes with the quality of teachers Models for boys and girls: Social, school, household covariates affect boys and girls differently: eg Household income, size, school fees Conclusion: Gender really matters in school dropout both from the micro and macro prospects [...]... handicrafts increases the occurrence of school dropout 2 The probability of school dropout varied with economic level of commune/ward School context: 1 The pyramid structure of school system results in a strict selective process in children’s schooling 2 The probability of school dropout increases with the school fees Interaction term: 1 The interaction between gender and socioeconomic level of commune indicates... indicates that a gender difference in school dropout is more likely to appear in the undeveloped areas, narrows with the increase of human development level, and disappears in the highly developed areas 2 The significant interaction between gender and the quality of teachers indicates that the gender gap in school dropout changes with the quality of teachers Models for boys and girls: Social, school, household...•Household level 1 The odds of school dropout decrease with the household income and the proportion of household expenditures on education 2 The probability increases with the number of siblings in the household, but decreases with the household size 3 The household head’s education level is positively related to children staying in school Commune/ward context: 1 Whether the commune/ward... school dropout changes with the quality of teachers Models for boys and girls: Social, school, household covariates affect boys and girls differently: eg Household income, size, school fees Conclusion: Gender really matters in school dropout both from the micro and macro prospects