This thesis uses log-linear models to analyze data from the 2000 China Population Census and examine the regional variation in educational assortative marriage in China. My results reveal considerable regional variation at the macro-regional, regional, and provincial levels, which challenges the common assumption of a national uniform marriage market. Inconsistent estimates of different assumptions of marriage market may be due to the geographic variation of educational assortative marriage or different educational distribution by region. I further study the geographic variation of educational assortative marriage and find that only the tendency of educational heterogamy shows substantial differences by province. These provincial differences in the tendency of educational hete...[ Read more ]

This thesis uses log-linear models to analyze data from the 2000 China Population Census and examine the regional variation in educational assortative marriage in China. My results reveal considerable regional variation at the macro-regional, regional, and provincial levels, which challenges the common assumption of a national uniform marriage market. Inconsistent estimates of different assumptions of marriage market may be due to the geographic variation of educational assortative marriage or different educational distribution by region. I further study the geographic variation of educational assortative marriage and find that only the tendency of educational heterogamy shows substantial differences by province. These provincial differences in the tendency of educational heterogamy show obvious geographic patterns, especially for the strength of educational hypergamy. The provinces in Middle China exhibit the strongest tendency for educational hypergamy, while the provinces in North China show the weakest such tendency. Geographic variation in educational hypergamy implies that when we compare the temporal trend of educational assortative marriage, it’s better to use the assumption of a local marriage market and include the geographic variation in the analysis.