Skip to main content
. Author manuscript; available in PMC: 2017 Aug 1.
Published in final edited form as: Sociol Methodol. 2016 Sep 20;46(1):84–120. doi: 10.1177/0081175015599808

Table 5.

Rural-urban stratified coefficient estimates from standard ordered probit versus hierarchical ordered probit (HOPIT) models of self-rated health: CFPS-2012.

Rural Sample
(N = 16,002)
Urban Sample
(N = 7,205)
Ordered
Probit
HOPIT
Ordered Probit HOPIT
Parallel Shift Non-Parallel
Shift
Parallel Shift Non-Parallel
Shift
Age (mean-centered and divided by 10) −0.285*** −0.304*** −0.312*** −0.261*** −0.261*** −0.265***
Age-square 0.022*** 0.004 0.005 0.057*** 0.048*** 0.042***
Male (ref: female) 0.238*** 0.310*** 0.306*** 0.204*** 0.303*** 0.296***
Marital status (ref: married/cohabit)
    Single −0.034 −0.054 −0.034 −0.072 −0.218*** −0.199**
    Divorced/Widowed −0.065 −0.046 −0.033 −0.102 −0.096 −0.067
Years of education 0.003 0.011*** 0.011*** 0.009* 0.013** 0.010*
Episodic memory 0.032*** 0.020** 0.019** 0.046*** 0.017 0.015
Employed (ref: no) 0.164*** 0.163*** 0.157*** 0.165*** 0.206*** 0.210***
Family income quartiles (poorest)
    2nd −0.005 −0.004 0.000 0.024 −0.045 −0.041
    3rd 0.017 0.068* 0.073** 0.064 0.000 0.008
    4th (richest) 0.007 0.044 0.044 0.091* 0.035 0.038
Cadre/Party membership (ref: no) 0.098* 0.001 0.000 0.049 0.008 0.007
Family cadre/Party member (ref: no) 0.041 0.032 0.039 0.096** 0.078* 0.076*
Urban hukou (ref: rural hukou) −0.020 −0.033 −0.026 −0.083* −0.117** −0.101*
Σ 1.284*** 1.271*** 1.154*** 1.157***

Notes: Estimates of ancillary parameters are not shown. All the models control for regional variation. Coefficients in the HOPIT models have been rescaled to be comparable to those in the ordered probit model.

*

p<0.05;

**

p<0.01;

***

p<0.001.