# Table 2 Income effects of occupational interaction based on the benchmark difference model N = 6263

Variables Least square method model First-order difference model
M1 M2 M3
Explanatory variable Monthly income of first job Monthly income of present job Monthly income difference
Internal interaction −.005 .010 .012*
(.004) (.008) (.004)
Outward interaction .051*** .057** .031**
(.008) (.013) (.007)
Seeking job by pulling strings .234*** .012 .053
(.041) (.034) (.032)
Male −.084** −.139**
(.023) (.032)
Party member .431*** .152** .345***
(.049) (.034) (.044)
Frequency of job changes .018 .023
(.009) (.013)
Intercept .304*** .874*** .664***
(.050) (.083) (.050)
R square .094 .092 .033
Sample size 6263 6263 6263
1. Note: (1) *p < 0.05, **p < 0.01, ***p < 0.001. (2) Values inside the parentheses are standard errors obtained by taking heteroskedasticity robustness and urban cluster robustness into account. (3) In a difference mode, what an explanatory variable corresponds to is the change value. For the conciseness of the table, different numerical values of dummy variables such as unit nature, way of gaining entry to work and occupation, etc. are processed as continuous variables. In a logical sense, their establishments do not signify specific meanings. For example, variables for the way of gaining entry to work are −1, 0, and 1, standing for variation from pulling strings to not pulling strings, being invariant, and from not pulling strings to pulling strings, separately. Therefore, it represents the degree of “changing as pulling strings” in practice. In robustness analysis, original dummy variables are adopted to perform tests and highly consistent results are achieved