Page 19 - IJPS-4-1

Xiaorong Gu

partially supports the inverted U-curve relationship theory but disputes the argument that educational homogamy is
stronger in Confucian societies. With data from 10 Asian societies, Smits and Park (2009) revealed that, rather than
an inverted U-curve relationship, there is a positive linear relationship between educational homogamy and economic
modernization.
Scholars report mixed findings on the homogamy patterns in China. For example, Xu, Ji, and Tung (2000) examined
the homogamy patterns across three marriage cohorts in two cities, revealing persistent salience of homogamy and trade-
off between social and political status characteristics in mate selection. Song’s (2009) analysis of a representative data of
urban residents, however, shows significant evidence of weaker homogamy among the Cultural Revolutionary marriage
cohort. However, Han’ s (2010) analysis of the 2000 China Population Census data indicates that, since the early 1980s,
homogamy rates have seen steady growths for two decades, albeit fluctuations in the rural sample. This is supported
by Xu, Li, and Yu (2014) finding based on China Family Panel Studies data, which documents increasing educational
homogamy in China. More recent evidence reveals more nuances if accounting for the effects of gender, educational
level, and hukou status. According to Qian and Qian (2014), the increasing educational attainment of urban females has a
squeeze effect on their marital prospects: While marriage rates for females who received less than college education are
higher than those for males, college-educated females have lower marriage rates than their male counterparts.
Below I identify two research gaps in existing literature. First, these two dimensions of the topic, i.e., the timing
of marriage and the homogamy patterns could and should be incorporated in a systematic study of youths’ marriage
behaviors, rather than in two separate bodies of literature. For one thing, considering the relatively low rates of divorce
and cohabitation which largely functions as a transition to legal marriage (Yu and Xie, 2015a) in China, I argued that
the homogamy patterns of first marriages could approximate the general conditions of all marriages. That is, to study
homogamy patterns of first marriages among young adults lends important insights to our understanding of family
formation and social stratification in China at large. For another, from the perspective of youth studies, knowledge of
both young people’s marriage timing and the patterns of their spouse selection yields a holistic picture of the changes and
continuities in the institution of marriage for generations of youths, which has ample theoretical and practical implications.
Second, as described above, while many previous projects employed cohort study as the method to investigate how
sociopolitical contexts affect individual life, in the study of youths’ transition to adulthood after reform, with a few
exceptions (e.g., Yeung and Hu, 2013), rarely do researchers distinguish different stages of economic reform. With the
velocity of social change in post-reform China, lumping all those who come of age after 1978 together could be analytically
erroneous. Therefore, a more fine-grained cohort schema to capture different stages of reform is recommended. In this
analysis, I drew on three repeated cross-sectional data from CGSS to investigate the marriage formation of four cohorts
of young adults in different historical contexts. Specifically, I addressed two major research questions: (1) What are the
patterns of marriage timing among Chinese youths over different birth cohorts? And (2) What are the trends of assortative
marriage in across different historical periods?
4. Data and Methods

4.1. Data and sample
To address the research questions above, I analyze pooled data from the 2012, 2013, and 2015 waves of the CGSS, which
is a national representative household survey with stratified samples of respondents aged 18–69, drawn from 31 provinces/
districts (excluding Taiwan, Hong Kong, and Macau) (see details in CGSS, 2010). This second cycle is preceded by an
earlier cycle between 2003 and 2008 which yielded four waves of data publicly available. Surveys conducted in this cycle
cover about 12,000 households in each wave, collecting comprehensive information on respondents’ life history, family
composition, education, social attitudes, employment conditions, and social networks, hence allowing systematic analysis
of marriage patterns among subgroups across birth cohorts.
Since each of the CGSS 2012, 2013, and 2015 datasets includes detailed information on respondents’ life history
and marriage behaviors, I pooled these data to systematically examine the patterns of Chinese youths’ transition to first
marriage. I used the survey year as a dummy variable in the Cox hazard model to control for potential effects of the
interview time on the outcome variables. To analyze youths’ first marriage timing, I included 27,887 respondents (9,760
from the 2012 wave, 9,202 from the 2013 wave, and 8,925 from the 2015 wave), who were born in 1946–1993 and were at
least 22 years old at interview, which is the legal age threshold for all Chinese citizens since 1980 (The National People’s
Congress, 1980). I excluded an earlier cohort born in 1936–1945 in the analysis due to potential sample selectivity resulted
from high mortality rates in elderly populations (Lindenberger, Singer, and Baltes, 2002). As such, the sample covers

International Journal of Population Studies | 2018, Volume 4, Issue 1 13