Hello everyone! I am new here and am very happy to come to STATALIST!
I am working on my paper with cross section data of CFPS2014(China Family Panel Studies).The dependent variable (satis14) is an ordial variable(level 1-4); and the independent variable (fam_travel14) is dummy variable and endogenous.; furthermore I use “tourist_attractions”as instrumental variable.Because the dependent variable and independent variable cause and effect each other cause and effect each other, and they are both discrete variables, the two-stage regression method based on continuous variables is not suitable. I tried to use CMP(Conditional mixed process) and Bioprobit model after doing OLS and Oprobit. Unfortunately, both CMP and Bioprobit commands yield different results from OLS and Oprobit. There are remarkable positive relevant relations between the dependent variable and independent variable by OLS and Oprobit; but there are remarkable negative relevant relations between the dependent variable and independent variable by CMP and Bioprobit, and there are remarkable negative relevant relations between independent variable and instrumental variable, which is illogical in the actual life.
Can anybody help me analyze where the problem is? Thank you very much!

The OLS command and results is:
Code:
satis14 fam_travel14 gender14 age14 age14_square eduyears14 hukou14 marriage14 health14 religion14 employ14 job_nature14 job_type14 par_trust14 n_fmeals14 familysize14 child_num14 residence14 house14 logfincome14 fam_med14 lfam_med14 f_debts14 lfam_debts14 f_hdebts14 lhouse_debts14 f_fdebts14 lnonhousing_debts14,robust

The Oprobit command is:
Code:
oprobit satis14 fam_travel14 gender14 age14 age14_square eduyears14 hukou14 marriage14 health14 religion14 employ14 job_nature14 job_type14 par_trust14 n_fmeals14 familysize14 child_num14 residence14 house14 logfincome14 fam_med14 lfam_med14 f_debts14 lfam_debts14 f_hdebts14 lhouse_debts14 f_fdebts14 lnonhousing_debts14,robust


The Bioprobit command is:
Code:
bioprobit(fam_travel14=gender14 age14 age14_square eduyears14 hukou14 marriage14 health14 religion14 employ14 job_nature14 job_type14 par_trust14 n_fmeals14 sstatus14 familysize14 child_num14 residence14 house14 logfincome14 fam_med14 lfam_med14 f_debts14 tourist_attractions)  (satis14=fam_travel14 gender14 age14 age14_square eduyears14 hukou14 marriage14 health14 religion14 employ14 job_nature14 job_type14 par_trust14 n_fmeals14 sstatus14 familysize14 child_num14 residence14 house14 logfincome14 fam_med14 lfam_med14 f_debts14)


The CMP command is:
Code:
cmp(satis=fam_travel14 gender14 age14 age14_square eduyears14 hukou14 marriage14 health14 religion14 employ14 job_nature14 job_type14  par_trust14 n_fmeals14 sstatus14 familysize14 child_num14 residence14 house14 logfincome14 fam_med14 lfam_med14 f_debts14)  (fam_travel14=tourist_attractions gender14 age14 age14_square eduyears14 hukou14 marriage14 health14 religion14 employ14 job_nature14 job_type14 par_trust14 n_fmeals14 sstatus14 familysize14 child_num14 residence14  house14 logfincome14 fam_med14 lfam_med14 f_debts14),ind($cmp_probit $cmp_oprobit) tech(dfp) nolr quietly robust