Page 107 - IJPS-1-1

Kuang-Chi Chang

Table 1. Sample statistics of migrants and urban natives in Shanghai 2008
Mean/Percentage
Migrants (N=334) Urban Natives (N=540)
Dependent Variable:
Job search channels
Market 27.85%*** 17.90%
Hierarchy 16.78%*** 53.70%
Network 55.37%*** 28.40%
Independent Variable:
Job search period
Before 1990 2.30%*** 21.50%
Between 1990 and 1999 15.08%*** 19.47%
Between 2000 and 2004 31.80%*** 25.96%
2005 and after 50.82%*** 33.06%
Control Variables:
Age 35.52 42.03
Male 45.81%*** 53.70%
Education
Elementary and below 14.97%*** 2.24%
Middle school 66.17%*** 25.61%
High school 18.86%*** 40.93%
College and above 0%*** 31.21%
Workplace
Private sector 85.48%*** 37.45%
State sector 14.52%*** 62.55%
t-tests are used to examine the differences between migrants and urban natives for continuous variables. Chi-square tests are used for categorical variables.
* p<0.10, ** p<0.05, *** p<0.01.

the dependent variable job search channel (market, hierarchy and network, respectively).
Since multinomial logistic regression is used to predict probability of choosing one out-
come category (a job search channel) over the probability of choosing the baseline cate-
gory (another job search channel) based on multiple independent variables, the odds ratios
differ depending on the base category specified, even though the underlying model is the
same (hence the same summary statistics). Using market method as the base category, the
first set of results shows that, compared to urban natives, rural migrants are less likely to
use hierarchy (odds ratio=0.51; p<0.05) over market method in finding jobs. Although the
proportion t-tests find the differences between the proportions of using market versus net-
work methods are statistically significant for both rural migrants and urban natives, the
regression results are inconclusive to these differences. There is no sufficient evidence to
support Hypotheses 1.2 and 2.2 that migrants and urban natives (respectively) are more
likely to use network than market channels. Results also do not find significant effects on
the likelihood of rural migrants versus urban natives for their reliance on network over
market method, thus supporting Hypotheses 3.1 and 3.2 which expect rural migrants and
urban natives to not differ in their reliance on network and on market method in their job
search.
Compared to individuals who found their current or last jobs after 2005, those who
found their jobs during the 1980s and the 1990s were more likely to use hierarchy than

International Journal of Population Studies | 2015, Volume 1, Issue 1 101