******************************************************************************
************************IMPUTING MISSING OBS**********************************
******************************************************************************

//0. GET THE DATA SET WE NEED: MERGE
use "D:\Barbara Castillo Rico\2. Project\2. Data\1. Elfe\3. Data set\FINALbase160211.dta", clear //latest


//1. WE NEED A VARIABLE TO IDENTIFY THE CHILDREN WE KEEP IN OUR ANALYSIS
* Children for which we have data in 2months wave
g months2 = 1 if inlist(m02x_qmerecomp2m,1,2) & inlist(m02x_typqmere2m,1,2,3) //completed and partially completed, mother referent
replace months2 = 1 if inlist(m02x_qpecfcomp2m,1,2) & inlist(m02x_typqpecf2m,1,2) //completed and partially completed, father referent
replace months2 = 1 if m02x_partimere2m==10 & missing(months2)
replace months2 = 0 if missing(months2)
label var months2 "the child is included in the 2months wave"
label define months2 1 "yes" 0 "no"
label values months2 months2
tab months2, missing


//2. we need to identify fathers who are not present in elfe
g father_recogn_child=.
replace father_recogn_child = 1 if m00m2_etatmat==1 & m00m2_couple==1
replace father_recogn_child = 1 if m00m2_reconu==1 & father_recogn_child==.
replace father_recogn_child = 1 if m02m_recon==1 & father_recogn_child==.
replace father_recogn_child = 0 if m00m2_reconu==0 & father_recogn_child==.
replace father_recogn_child = 0 if m02m_recon==2 & father_recogn_child==.

label var father_recogn_child "father has recognized the child, 2 months"
label define father_recogn_child 0 "no" 1 "yes" 
tab father_recogn_child, missing
*
g child_hhld = 1 if m02m_efvit==1
replace child_hhld = 2 if m02m_efvit==2
replace child_hhld = 3 if m02m_efvit==3
replace child_hhld = 4 if m02m_efvit==4
replace child_hhld = 5 if m02m_efvit==5
replace child_hhld = 6 if inlist(m02m_efvit,6,7)

replace child_hhld = 6 if m02x_typqmere2m==3 & missing(child_hhld)
replace child_hhld = 5 if m02x_typqmere2m==4 & missing(child_hhld)

replace child_hhld = 3 if m02x_typqpecf2m==3 & missing(child_hhld)
replace child_hhld = 6 if m02x_typqpecf2m==5 & missing(child_hhld)
replace child_hhld = 5 if m02x_typqpecf2m==6 & missing(child_hhld)
replace child_hhld = 2 if m02x_typqpecf2m==7 & missing(child_hhld)
replace child_hhld = 6 if m02x_typqpecf2m==8 & missing(child_hhld)
replace child_hhld = 5 if m02x_typqpecf2m==9 & missing(child_hhld)

replace child_hhld = 1 if m02m_lientyp_4==1 & missing(child_hhld)
replace child_hhld = 2 if !inlist(m02m_lientyp_4,1) & missing(child_hhld)

label var child_hhld "person living with the child, 2 months"
label define child_hhld 1 "both father and mother" 2 "mother" 3 "father" 4 ///
"rotating custody" 5 "placed child" 6 "at the hospital"
tab child_hhld, missing

**if father is not in the household, the father did not recognise the child, //
* and he did not answer the questionnaire....
g f_unknown = 1 if child_hhld==2 & !inlist(m02p_lientyp_4,1) & ///
father_recogn_child==0 
replace f_unknown = 1 if inlist(m02m_infper,2,3,4) & f_unknown==.

replace f_unknown = 0 if f_unknown==.

lab var f_unknown "the father is unknown to elfe, 2 months"
lab define f_unknown 1 "yes" 0 "no"
lab values f_unknown f_unknown
tab f_unknown, missing
tab f_unknown if months2==1, missing






*******************************************************************************
************************** FATHER'S FAMILY ************************************
*******************************************************************************


//3. INFORMATION ABOUT FATHER'S FAMILY

//3.1. Father
//Born in france
g fbirthfr = 1 if m02p_lnais_4==1 & m02p_lientyp_4==1
replace fbirthfr = 0 if m02p_lnais_4==2 & m02p_lientyp_4==1
replace fbirthfr = 1 if m02m_lnais_4==1 & m02m_lientyp_4==1 & missing(fbirthfr)
replace fbirthfr = 0 if m02m_lnais_4==2 & m02m_lientyp_4==1 & missing(fbirthfr)
replace fbirthfr = 1 if m00m2_lieunaisp==1 & missing(fbirthfr)
replace fbirthfr = 0 if m00m2_lieunaisp==2 & missing(fbirthfr)

label var fbirthfr "father born in france"
label define fbirthfr 1 "yes" 0 "no"
label values fbirthfr fbirthfr
tab fbirthfr, missing

//French nationality at birth
g fnationfr = 1 if m02p_natio1n_4==1 & m02p_lientyp_4==1 //born with french nationality
replace fnationfr = 0 if inlist(m02p_natio1n_4,2,3,4) & m02p_lientyp_4==1
replace fnationfr = 1 if m02m_natio1n_4==1 & m02m_lientyp_4==1 & missing(fnationfr)
replace fnationfr = 0 if inlist(m02m_natio1n_4,2,3,4) & m02m_lientyp_4==1 & missing(fnationfr)
replace fnationfr = 1 if m00m2_natiop==1 & missing(fnationfr)
replace fnationfr = 0 if inlist(m00m2_natiop,2,3) & missing(fnationfr)

label var fnationfr "father french nationality at birth"
label define fnationfr 1 "yes" 0 "no"
label values fnationfr fnationfr
tab fnationfr, missing

//father's lenguage
g flenghome = 1 if m02p_lang1e==2 //lenguage spoken at home with the child
replace flenghome = 0 if flenghome==. & !inlist(m02p_lang1e,.)
label var flenghome "lenguage spoken at home with the child, father"
label define flenghome 1 "french" 0 "other"

g flengent = 1 if m02p_lang==1 //interview lenguage
replace flengent = 0 if flengent==. & !inlist(m02p_lang,.)
label var flengent "lenguage at interview, father"
label define flengent 1 "french" 0 "other"

g fleng = 1 if flenghome==1 & flengent==1 
replace fleng = 0 if flenghome==0 & flengent==1 //french origin lenguage
replace fleng = 0 if flenghome==0 & flengent==0 //foreign lenguage as origin
replace fleng = 2 if flenghome==1 & flengent==0 //french at home but foreign lenguage at interview
label var fleng "main lenguage"
label define fleng 1 "french" 0 "foreign lenguage" 2 "not clear"
tab fleng, missing


//3.2. Grandmother father
//Born in france
g f_gmbirthfr = 1 if m02p_mblieu==1
replace f_gmbirthfr = 0 if m02p_mblieu==2
foreach i of numlist 5/10{ //5/11
replace f_gmbirthfr = 1 if m02p_lnais_`i'==1 & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==2 & missing(f_gmbirthfr) //grandparent (father), female, born in france
replace f_gmbirthfr = 0 if m02p_lnais_`i'==2 & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==2 & missing(f_gmbirthfr)
}
replace f_gmbirthfr = 1 if m02m_bmlieu==1 & missing(f_gmbirthfr)
replace f_gmbirthfr = 0 if m02m_bmlieu==2 & missing(f_gmbirthfr)
foreach i of numlist 4/10{
replace f_gmbirthfr = 1 if m02m_lnais_`i'==1 & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==2 & missing(f_gmbirthfr) //grandparent (father), female, born in france
replace f_gmbirthfr = 0 if m02m_lnais_`i'==2 & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==2 & missing(f_gmbirthfr)
}

label var f_gmbirthfr "grandmother born in france (father)"
label define f_gmbirthfr 1 "yes" 0 "no"
label values f_gmbirthfr f_gmbirthfr
tab f_gmbirthfr, missing

//French nationality at birth
g f_gmnationfr = 1 if m02p_mbfranc==1
replace f_gmnationfr = 0 if inlist(m02p_mbfranc,2,3,4)
foreach i of numlist 5/10{ //5/11
replace f_gmnationfr = 1 if m02p_natio1n_`i'==1 & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==2 & missing(f_gmnationfr) //grandparent (mother), female french nationality at birth
replace f_gmnationfr = 0 if inlist(m02p_natio1n_`i',2,3,4) & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==2 & missing(f_gmnationfr)
}
replace f_gmnationfr = 1 if m02m_bmfranc==1 & missing(f_gmnationfr)
replace f_gmnationfr = 0 if inlist(m02m_bmfranc,2,3,4) & missing(f_gmnationfr)
foreach i of numlist 4/10{
replace f_gmnationfr = 1 if m02m_natio1n_`i'==1 & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==2 & missing(f_gmnationfr) //grandparent (mother), female french nationality at birth
replace f_gmnationfr = 0 if inlist(m02m_natio1n_`i',2,3,4) & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==2 & missing(f_gmnationfr)
}

label var f_gmnationfr "grandmother french nationality at birth (father)"
label define f_gmnationfr 1 "yes" 0 "no"
label values f_gmnationfr f_gmnationfr
tab f_gmnationfr, missing

//grandmother father main lenguage
g f_gmleng = 1 if m02p_peremlan==1
replace f_gmleng = 0 if f_gmleng==. & !inlist(m02p_peremlan,.)
replace f_gmleng = 1 if m02p_mblang==1 & f_gmleng==.
replace f_gmleng = 0 if !inlist(m02p_mblang,.) & f_gmleng==.
replace f_gmleng = 1 if m02m_peremlan==1 & f_gmleng==.
replace f_gmleng = 0 if !inlist(m02m_peremlan,.) & f_gmleng==.
replace f_gmleng = 1 if m02m_bmlang==1 & f_gmleng==. & m02m_lientyp_4==1
replace f_gmleng = 0 if !inlist(m02m_bmlang,.) & !inlist(m02m_bmlang,1) & !inlist(m02m_bmlang,999) & m02m_lientyp_4==1 //not condition
//on missing data, because there are some differences and i tend to trust info on foreign lenguage

label var f_gmleng "main langue grand-mère"
label define f_gmleng 1 "français" 0 "autre langue"
tab f_gmleng


//3.3. Grandfather father
//Born in france
g f_gfbirthfr = 1 if m02p_pblieu==1
replace f_gfbirthfr = 0 if m02p_pblieu==2
foreach i of numlist 5/10{ //5/11
replace f_gfbirthfr = 1 if m02p_lnais_`i'==1 & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==1 & missing(f_gfbirthfr) //grandparent (father), male, born in france
replace f_gfbirthfr = 0 if m02p_lnais_`i'==2 & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==1 & missing(f_gfbirthfr)
}
replace f_gfbirthfr = 1 if m02m_bplieu==1 & missing(f_gfbirthfr)
replace f_gfbirthfr = 0 if m02m_bplieu==2 & missing(f_gfbirthfr)
foreach i of numlist 4/10{
replace f_gfbirthfr = 1 if m02m_lnais_`i'==1 & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==1 & missing(f_gfbirthfr) //grandparent (father), male, born in france
replace f_gfbirthfr = 0 if m02m_lnais_`i'==2 & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==1 & missing(f_gfbirthfr)
}

label var f_gfbirthfr "grandfather born in france (father)"
label define f_gfbirthfr 1 "yes" 0 "no"
label values f_gfbirthfr f_gfbirthfr
tab f_gfbirthfr, missing


//French nationality at birth
g f_gfnationfr = 1 if m02p_pbfranc==1
replace f_gfnationfr = 0 if inlist(m02p_pbfranc,2,3,4)
foreach i of numlist 5/10{ //5/11
replace f_gfnationfr = 1 if m02p_natio1n_`i'==1 & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==1 & missing(f_gfnationfr) //grandparent (father), male french nationality at birth
replace f_gfnationfr = 0 if inlist(m02p_natio1n_`i',2,3,4) & m02p_lientyp_`i'==9 & m02p_sexec1_`i'==1 & missing(f_gfnationfr)
}
replace f_gfnationfr = 1 if m02m_bpfranc==1 & missing(f_gfnationfr)
replace f_gfnationfr = 0 if inlist(m02m_bpfranc,2,3,4) & missing(f_gfnationfr)
foreach i of numlist 4/10{
replace f_gfnationfr = 1 if m02m_natio1n_`i'==1 & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==1 & missing(f_gfnationfr) //grandparent (father), male french nationality at birth
replace f_gfnationfr = 0 if inlist(m02m_natio1n_`i',2,3,4) & m02m_lientyp_`i'==9 & m02m_sexec1_`i'==1 & missing(f_gfnationfr)
}
label var f_gfnationfr "grandfather french nationality at birth (father)"
label define f_gfnationfr 1 "yes" 0 "no"
label values f_gfnationfr f_gfnationfr
tab f_gfnationfr, missing

//grandfather father main lenguage
g f_gfleng = 1 if m02p_pereplan==1
replace f_gfleng = 0 if f_gfleng==. & !inlist(m02p_pereplan,.)
replace f_gfleng = 1 if m02p_pblang==1 & f_gfleng==.
replace f_gfleng = 0 if !inlist(m02p_pblang,.) & f_gfleng==.
replace f_gfleng = 1 if m02m_pereplan==1 & f_gfleng==.
replace f_gfleng = 0 if !inlist(m02m_pereplan,.) & f_gfleng==.
replace f_gfleng = 1 if m02m_bplang==1 & f_gfleng==. & m02m_lientyp_4==1
replace f_gfleng = 0 if !inlist(m02m_bplang,.,1,999) & m02m_lientyp_4==1 //not condition
//on missing data, because there are some differences and i tend to trust info on foreign lenguage

label var f_gfleng "main lenguage grandfather father"
label define f_gfleng 1 "french" 0 "other"
tab f_gfleng



//4. IMPUTATIONS OF FATHER' FAMILLY -> LENGUAGE
//4.1. FATHER
tab fleng fbirthfr, missing
tab fleng fbirthfr //[aweight = M00E_PONDVAL]
replace fbirthfr = rbinomial(1, 0.93) if fbirthfr==. & fleng==1
tab fleng fbirthfr, missing

tab fleng fnationfr, missing
tab fleng fnationfr //[aweight = M00E_PONDVAL]
replace fnationfr = rbinomial(1, 0.245) if fnationfr==. & flenghome==0
replace fnationfr = rbinomial(1, 0.936) if fnationfr==. & flenghome==1
tab fleng fnationfr, missing

//4.2. GRANDMOTHER FATHER
tab f_gmleng f_gmbirthfr, missing
tab f_gmleng f_gmbirthfr
replace f_gmbirthfr = rbinomial(1, 0.164) if f_gmleng==0 & f_gmbirthfr==.
replace f_gmbirthfr = rbinomial(1, 0.902) if f_gmleng==1 & f_gmbirthfr==.
tab f_gmleng f_gmbirthfr, missing

tab f_gmleng f_gmnationfr, missing
tab f_gmleng f_gmnationfr
replace f_gmnationfr = rbinomial(1, 0.195) if f_gmleng==0 & f_gmnationfr==.
replace f_gmnationfr = rbinomial(1, 0.937) if f_gmleng==1 & f_gmnationfr==.
tab f_gmleng f_gmnationfr, missing

//4.3. GRANDFATHER FATHER
tab f_gfleng f_gfbirthfr, missing
tab f_gfleng f_gfbirthfr
replace f_gfbirthfr = rbinomial(1, 0.171) if f_gfleng==0 & f_gfbirthfr==.
replace f_gfbirthfr = rbinomial(1, 0.882) if f_gfleng==1 & f_gfbirthfr==.
tab f_gfleng f_gfbirthfr, missing

tab f_gfleng f_gfnationfr, missing
tab f_gfleng f_gfnationfr
replace f_gfnationfr = rbinomial(1, 0.193) if f_gfleng==0 & f_gfnationfr==.
replace f_gfnationfr = rbinomial(1, 0.923) if f_gfleng==1 & f_gfnationfr==.
tab f_gfleng f_gfnationfr, missing



//5. IMPUTATIONS FATHER' FAMILLY -> OTHER MEMBERS' INFO

//5.1. From grandparents' information, FATHER
g f_parents = 1 if f_gfnationfr==1 | f_gmnationfr==1
replace f_parents = 0 if f_gfnationfr==0 & f_gmnationfr==0
label var f_parents "At least one grandparent has french nationality, father"
label define f_parents 1 "Yes" 0 "No"

tab fbirthfr fnationfr, missing
replace fnationfr = 1 if f_parents==1 & fnationfr==.
replace fnationfr = 0 if f_parents==0 & fnationfr==.
tab fbirthfr fnationfr, missing



//6. IMPUTATIONS FATHER' FAMILLY -> OWN BIRTH PLACE AND NATIONALITY

//6.1. FATHER
tab fbirthfr fnationfr, missing
tab fbirthfr fnationfr
replace fbirthfr = rbinomial(1, 0.05) if fnationfr==0 & fbirthfr==.
replace fbirthfr = rbinomial(1, 0.983) if fnationfr==1 & fbirthfr==.
replace fnationfr = rbinomial(1, 0.099) if fnationfr==. & fbirthfr==0
replace fnationfr = rbinomial(1, 0.992) if fnationfr==. & fbirthfr==1
tab fbirthfr fnationfr, missing




//7. IMPUTATIONS FATHER' FAMILLY -> OTHER MEMBERS' INFO

//7.1. From father nationality, GRANDMOTHER FATHER
tab f_gmbirthfr f_gmnationfr, missing
replace f_gmnationfr = 0 if fnationfr==0 & f_gmnationfr==.
replace f_gmnationfr = 1 if fnationfr==1 & f_gfnationfr==0 & f_gmnationfr==.
tab f_gmbirthfr f_gmnationfr, missing

//7.2. From father nationality, GRANDFATHER FATHER
tab f_gfbirthfr f_gfnationfr, missing
replace f_gfnationfr = 0 if fnationfr==0 & f_gfnationfr==.
replace f_gfnationfr = 1 if fnationfr==1 & f_gmnationfr==0 & f_gfnationfr==.
tab f_gfbirthfr f_gfnationfr, missing


//8. IMPUTATIONS FATHER' FAMILLY -> OWN BIRTH PLACE AND NATIONALITY

//8.1. GRANDMOTHER FATHER
tab f_gmbirthfr f_gmnationfr
replace f_gmbirthfr = rbinomial(1, 0.952) if f_gmnationfr==1 & f_gmbirthfr==.
replace f_gmbirthfr = rbinomial(1, 0.038) if f_gmnationfr==0 & f_gmbirthfr==.
replace f_gmnationfr = rbinomial(1, 0.993) if f_gmnationfr==. & f_gmbirthfr==1
replace f_gmnationfr = rbinomial(1, 0.21) if f_gmnationfr==. & f_gmbirthfr==0
tab f_gmbirthfr f_gmnationfr, missing

//8.2. GRANDFATHER FATHER
tab f_gfbirthfr f_gfnationfr
replace f_gfbirthfr = rbinomial(1, 0.947) if f_gfnationfr==1 & f_gfbirthfr==.
replace f_gfbirthfr = rbinomial(1, 0.035) if f_gfnationfr==0 & f_gfbirthfr==.
replace f_gfnationfr = rbinomial(1, 0.993) if f_gfnationfr==. & f_gfbirthfr==1
replace f_gfnationfr = rbinomial(1, 0.215) if f_gfnationfr==. & f_gfbirthfr==0
tab f_gfbirthfr f_gfnationfr, missing



*********************** MIGRATION VARIABLES FATHER *****************************


//1. Grandparents father
g f_gmimm = 1 if f_gmbirthfr==0 & f_gmnationfr==0
replace f_gmimm = 0 if f_gmimm==. & !inlist(f_gmbirthfr,.) & !inlist(f_gmnationfr,.)
label var f_gmimm "Grandmother born abroad with foreign nationality (regardless of current residence place), father"
label define f_gmimm 1 "yes"
tab f_gmimm if months2==1 & f_unknow==0, missing

g f_gfimm = 1 if f_gfbirthfr==0 & f_gfnationfr==0
replace f_gfimm = 0 if f_gfimm==. & !inlist(f_gfbirthfr,.) & !inlist(f_gfnationfr,.)
label var f_gfimm "Grandfather born abroad with foreign nationality (regardless of current residence place), father"
label define f_gfimm 1 "yes"
tab f_gfimm if months2==1 & f_unknow==0, missing

**EXTRA IMPUTATION
tab f_gmimm f_gfimm if months==1 & f_unknow==0, missing //WE CAN IMPUTE MORE
tab f_gmimm f_gfimm, missing
replace f_gmimm = rbinomial(1, 0.04) if f_gfimm==0 & f_gmimm==.
replace f_gfimm = rbinomial(1, 0.055) if f_gmimm==0 & f_gfimm==.
tab f_gmimm f_gfimm if months==1 & f_unknow==0, missing


//2. Father
g fimm = 1 if fbirthfr==0 & fnationfr==0 //immigrant mother
replace fimm = 2 if f_gmimm==1 & f_gfimm==1 & fbirthfr==1 & missing(fimm) //mother descendant of two immigrants

g f_oneimm = 1 if f_gmimm==1 & f_gfimm==0
replace f_oneimm = 1 if f_gmimm==1 & f_gfimm==.
replace f_oneimm = 1 if f_gmimm==0 & f_gfimm==1
replace f_oneimm = 1 if f_gmimm==. & f_gfimm==1

replace fimm = 3 if f_oneimm==1 & fbirthfr==1 & missing(fimm) //mother descendant of at least one immigrant
replace fimm = 4 if !inlist(fbirthfr,.) & !inlist(fnationfr,.)& missing(fimm) //rest of population

label var fimm "father immigrant history"
label define fimm 1 "immigrant father" 2 "father descendant of two immigrants" 3 "father descendant of at least one immigrant" 4 "rest of population"
label values fimm fimm
tab fimm, missing
tab fimm if months2==1 & f_unknow==0, missing
tab fimm if months2==1 & f_unknow==0 & !missing(f_gfimm), missing


*****EXTRA
//9. WHAT ABOUT ALL MISSING OBSERVATIONS??
//FATHER
tab fbirthfr fnationfr, missing
tab fbirthfr fnationfr if months2==1, missing
tab fbirthfr fnationfr if f_unknown==0, missing
tab fbirthfr fnationfr if months2==1 & f_unknown==0, missing

**-> Excluding unknown fathers and keeping just children in 2 months survey, 
*we just have 28 missing values
*keep if fbirthfr==. & fnationfr==. & months2==1 & f_unknown==0
*tab M00M2_COUPLE M00M2_ETATMAT, missing
*tab M00M2_COUPLE child_hhld, missing
*tab child_hhld , missing

*keep M00M2_ETATMAT M00M2_COUPLE child_hhld M02M_LIENTYP_4 M02P_LIENTYP_4 ///
*M02X_CONTPECF2M father_recogn_child father_present_birth M02M_INFPER

//GRANDMOTHER, FATHER
tab f_gmbirthfr f_gmnationfr, missing
tab f_gmbirthfr f_gmnationfr if months2==1, missing
tab f_gmbirthfr f_gmnationfr if f_unknown==0, missing
tab f_gmbirthfr f_gmnationfr if months2==1 & f_unknown==0, missing

//GRANDFATHER, FATHER
tab f_gfbirthfr f_gfnationfr, missing
tab f_gfbirthfr f_gfnationfr if months2==1, missing
tab f_gfbirthfr f_gfnationfr if f_unknown==0, missing
tab f_gfbirthfr f_gfnationfr if months2==1 & f_unknown==0, missing
***








*******************************************************************************
************************** MOTHER'S FAMILY ************************************
*******************************************************************************

//3. INFORMATION ABOUT MOTHER'S FAMILY

//3.1. Mother
//Born in france, MOTHER
g mbirthfr = 1 if m02m_lnais_3==1 //born in france
replace mbirthfr = 0 if m02m_lnais_3==2 //born abroad
foreach i of numlist 3/10{
replace mbirthfr = 1 if m02p_lnais_`i'==1 & m02p_lientyp_`i'==2 & missing(mbirthfr)
replace mbirthfr = 0 if m02p_lnais_`i'==2 & m02p_lientyp_`i'==2 & missing(mbirthfr)
}
replace mbirthfr = 1 if m00m2_lieunaism==1 & missing(mbirthfr) 
replace mbirthfr = 0 if m00m2_lieunaism==2 & missing(mbirthfr) 

label var mbirthfr "mother born in france"
label define mbirthfr 1 "yes" 0 "no"
label values mbirthfr mbirthfr
tab mbirthfr, missing

//French nationality at birth, MOTHER
g mnationfr = 1 if m02m_natio1n_3==1 //born with french nationality
replace mnationfr = 0 if inlist(m02m_natio1n_3,2,3,4) //not born with french nationality
foreach i of numlist 3/10{
replace mnationfr = 1 if m02p_natio1n_`i'==1 & m02p_lientyp_`i'==2 & missing(mnationfr)
replace mnationfr = 0 if inlist(m02p_natio1n_`i',2,3,4) & m02p_lientyp_`i'==2 & missing(mnationfr)
}
replace mnationfr = 1 if m00m2_natiom==1 & missing(mnationfr) 
replace mnationfr = 0 if inlist(m00m2_natiom,2,3,4) & missing(mnationfr) 

label var mnationfr "mother french nationality at birth"
label define mnationfr 1 "yes" 0 "no"
label values mnationfr mnationfr
tab mnationfr, missing

//mother's lenguage, mother
g mlenghome = 1 if m02m_lang1e==2 //lenguage spoken at home with the child
replace mlenghome = 0 if mlenghome==. & !inlist(m02m_lang1e,.)
label var mlenghome "lenguage spoken at home with the child, mother"
label define mlenghome 1 "french" 0 "other"

g mlengent = 1 if m02m_lang==1 //interview lenguage
replace mlengent = 0 if mlengent==. & !inlist(m02m_lang,.)
label var mlengent "lenguage at interview, mother"
label define mlengent 1 "french" 0 "other"

g mleng = 1 if mlenghome==1 & mlengent==1 
replace mleng = 0 if mlenghome==0 & mlengent==1 //french origin lenguage
replace mleng = 0 if mlenghome==0 & mlengent==0 //foreign lenguage as origin
replace mleng = 2 if mlenghome==1 & mlengent==0 //french at home but foreign lenguage at interview
label var mleng "main lenguage, mother"
label define mleng 1 "french" 0 "foreign lenguage" 2 "not clear"
tab mleng, missing


//3.2. Grandmother mother
//Born in france
g m_gmbirthfr = 1 if m02m_mblieu==1
replace m_gmbirthfr = 0 if m02m_mblieu==2
foreach i of numlist 4/10{
replace m_gmbirthfr = 1 if m02m_lnais_`i'==1 & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==2 & missing(m_gmbirthfr) //grandparent (mother), female, born in france
replace m_gmbirthfr = 0 if m02m_lnais_`i'==2 & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==2 & missing(m_gmbirthfr)
}
replace m_gmbirthfr = 1 if m02p_bmlieu==1 & missing(m_gmbirthfr)
replace m_gmbirthfr = 0 if m02p_bmlieu==2 & missing(m_gmbirthfr)
foreach i of numlist 5/9{
replace m_gmbirthfr = 1 if m02p_lnais_`i'==1 & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==2 & missing(m_gmbirthfr) //grandparent (mother), female, born in france
replace m_gmbirthfr = 0 if m02p_lnais_`i'==2 & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==2 & missing(m_gmbirthfr)
}
label var m_gmbirthfr "grandmother born in france (mother)"
label define m_gmbirthfr 1 "yes" 0 "no"
label values m_gmbirthfr m_gmbirthfr
tab m_gmbirthfr, missing

//French nationality at birth
g m_gmnationfr = 1 if m02m_mbfranc==1
replace m_gmnationfr = 0 if inlist(m02m_mbfranc,2,3,4)
foreach i of numlist 4/10{
replace m_gmnationfr = 1 if m02m_natio1n_`i'==1 & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==2 & missing(m_gmnationfr) //grandparent (mother), female french nationality at birth
replace m_gmnationfr = 0 if inlist(m02m_natio1n_`i',2,3,4) & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==2 & missing(m_gmnationfr)
}
replace m_gmnationfr = 1 if m02p_bmfranc==1 & missing(m_gmnationfr)
replace m_gmnationfr = 0 if inlist(m02p_bmfranc,2,3,4) & missing(m_gmnationfr)
foreach i of numlist 5/9{
replace m_gmnationfr = 1 if m02p_natio1n_`i'==1 & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==2 & missing(m_gmnationfr) //grandparent (mother), female french nationality at birth
replace m_gmnationfr = 0 if inlist(m02p_natio1n_`i',2,3,4) & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==2 & missing(m_gmnationfr)
}
label var m_gmnationfr "grandmother french nationality at birth (mother)"
label define m_gmnationfr 1 "yes" 0 "no"
label values m_gmnationfr m_gmnationfr
tab m_gmnationfr, missing

//grandmother mother main lenguage
g m_gmleng = 1 if m02m_meremlan==1
replace m_gmleng = 0 if m_gmleng==. & !inlist(m02m_meremlan,.)
replace m_gmleng = 1 if m02m_mblang==1 & m_gmleng==.
replace m_gmleng = 0 if !inlist(m02m_mblang,.) & m_gmleng==.
replace m_gmleng = 1 if m02p_meremlan==1 & m_gmleng==.
replace m_gmleng = 0 if !inlist(m02p_meremlan,.) & m_gmleng==.
replace m_gmleng = 1 if m02p_bmlang==1 & m_gmleng==. & m02p_lientyp_4==1
replace m_gmleng = 0 if !inlist(m02p_bmlang,.) & !inlist(m02p_bmlang,1) & !inlist(m02p_bmlang,999) & m02p_lientyp_4==1 //not condition
//on missing data, because there are some differences and i tend to trust info on foreign lenguage

label var m_gmleng "main lenguage grandmother, mother"
label define m_gmleng 1 "french" 0 "other"
tab m_gmleng



//3.3. Grandfather
//Born in france
g m_gfbirthfr = 1 if m02m_pblieu==1
replace m_gfbirthfr = 0 if m02m_pblieu==2
foreach i of numlist 4/10{
replace m_gfbirthfr = 1 if m02m_lnais_`i'==1 & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==1 & missing(m_gfbirthfr) //grandparent (mother), male, born in france
replace m_gfbirthfr = 0 if m02m_lnais_`i'==2 & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==1 & missing(m_gfbirthfr)
}
replace m_gfbirthfr = 1 if m02p_bplieu==1 & missing(m_gfbirthfr)
replace m_gfbirthfr = 0 if m02p_bplieu==2 & missing(m_gfbirthfr)
foreach i of numlist 5/9{
replace m_gfbirthfr = 1 if m02p_lnais_`i'==1 & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==1 & missing(m_gfbirthfr) //grandparent (mother), male, born in france
replace m_gfbirthfr = 0 if m02p_lnais_`i'==2 & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==1 & missing(m_gfbirthfr)
}
label var m_gfbirthfr "grandfather born in france (mother)"
label define m_gfbirthfr 1 "yes" 0 "no"
label values m_gfbirthfr m_gfbirthfr
tab m_gfbirthfr, missing

//French nationality at birth
g m_gfnationfr = 1 if m02m_pbfranc==1
replace m_gfnationfr = 0 if inlist(m02m_pbfranc,2,3,4)
foreach i of numlist 4/10{
replace m_gfnationfr = 1 if m02m_natio1n_`i'==1 & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==1 & missing(m_gfnationfr) //grandparent (mother), male french nationality at birth
replace m_gfnationfr = 0 if inlist(m02m_natio1n_`i',2,3,4) & m02m_lientyp_`i'==8 & m02m_sexec1_`i'==1 & missing(m_gfnationfr)
}
replace m_gfnationfr = 1 if m02p_bpfranc==1 & missing(m_gfnationfr)
replace m_gfnationfr = 0 if inlist(m02p_bpfranc,2,3,4) & missing(m_gfnationfr)
foreach i of numlist 5/9{
replace m_gfnationfr = 1 if m02p_natio1n_`i'==1 & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==1 & missing(m_gfnationfr) //grandparent (mother), male french nationality at birth
replace m_gfnationfr = 0 if inlist(m02p_natio1n_`i',2,3,4) & m02p_lientyp_`i'==8 & m02p_sexec1_`i'==1 & missing(m_gfnationfr)
}
label var m_gfnationfr "grandfather french nationality at birth (mother)"
label define m_gfnationfr 1 "yes" 0 "no"
label values m_gfnationfr m_gfnationfr
tab m_gfnationfr, missing

//grandfather mother main lenguage
g m_gfleng = 1 if m02m_mereplan==1
replace m_gfleng = 0 if m_gfleng==. & !inlist(m02m_mereplan,.)
replace m_gfleng = 1 if m02m_pblang==1 & m_gfleng==.
replace m_gfleng = 0 if !inlist(m02m_pblang,.) & m_gfleng==.
replace m_gfleng = 1 if m02p_mereplan==1 & m_gfleng==.
replace m_gfleng = 0 if !inlist(m02p_mereplan,.) & m_gfleng==.
replace m_gfleng = 1 if m02p_bplang==1 & m_gfleng==. & m02p_lientyp_4==1
replace m_gfleng = 0 if !inlist(m02p_bplang,.,1,999) & m02p_lientyp_4==1 //not condition
//on missing data, because there are some differences and i tend to trust info on foreign lenguage

label var m_gfleng "main lenguage grandfather mother"
label define m_gfleng 1 "french" 0 "other"
tab m_gfleng



//4. IMPUTATIONS MOTHER'FAMILY -> LENGUAGE

//4.1. We keep initial proportions before any imputation for last step, MOTHER
tab mbirthfr mnationfr, missing
tab mbirthfr mnationfr

//4.2. MOTHER
tab mleng mbirthfr, missing
tab mleng mbirthfr //no missing obs

tab mleng mnationfr, missing
tab mleng mnationfr
replace mnationfr = rbinomial(1, 0.211) if mnationfr==. & mlenghome==0
replace mnationfr = rbinomial(1, 0.931) if mnationfr==. & mlenghome==1
tab mleng mnationfr, missing

//4.3. GRANDMOTHER MOTHER
tab m_gmleng m_gmbirthfr, missing
tab m_gmleng m_gmbirthfr
replace m_gmbirthfr = rbinomial(1, 0.142) if m_gmleng==0 & m_gmbirthfr==.
replace m_gmbirthfr = rbinomial(1, 0.889) if m_gmleng==1 & m_gmbirthfr==.
tab m_gmleng m_gmbirthfr, missing

tab m_gmleng m_gmnationfr, missing
tab m_gmleng m_gmnationfr
replace m_gmnationfr = rbinomial(1, 0.159) if m_gmleng==0 & m_gmnationfr==.
replace m_gmnationfr = rbinomial(1, 0.924) if m_gmleng==1 & m_gmnationfr==.
tab m_gmleng m_gmnationfr, missing

//4.4. GRANDFATHER MOTHER
tab m_gfleng m_gfbirthfr, missing
tab m_gfleng m_gfbirthfr
replace m_gfbirthfr = rbinomial(1, 0.144) if m_gfleng==0 & m_gfbirthfr==.
replace m_gfbirthfr = rbinomial(1, 0.871) if m_gfleng==1 & m_gfbirthfr==.
tab m_gfleng m_gfbirthfr, missing

tab m_gfleng m_gfnationfr, missing
tab m_gfleng m_gfnationfr
replace m_gfnationfr = rbinomial(1, 0.157) if m_gfleng==0 & m_gfnationfr==.
replace m_gfnationfr = rbinomial(1, 0.912) if m_gfleng==1 & m_gfnationfr==.
tab m_gfleng m_gfnationfr, missing


//5. IMPUTATIONS MOTHER'FAMILY -> OTHERS' INFO

//5.1. MOTHER
g m_parents = 1 if m_gfnationfr==1 | m_gmnationfr==1
replace m_parents = 0 if m_gfnationfr==0 & m_gmnationfr==0
label var m_parents "At least one grandparent has french nationality, mother"
label define m_parents 1 "Yes" 0 "No"

tab mbirthfr mnationfr, missing
replace mnationfr = 1 if m_parents==1 & mnationfr==.
replace mnationfr = 0 if m_parents==0 & mnationfr==.
tab mbirthfr mnationfr, missing


//6. IMPUTATIONS MOTHER'FAMILY -> BIRTH PLACE AND NATIONALITY

//6.1. MOTHER
tab mbirthfr mnationfr, missing
tab mbirthfr mnationfr
replace mbirthfr = rbinomial(1, 0.063) if mnationfr==0 & mbirthfr==.
replace mbirthfr = rbinomial(1, 0.986) if mnationfr==1 & mbirthfr==.
replace mnationfr = rbinomial(1, 0.094) if mnationfr==. & mbirthfr==0
replace mnationfr = rbinomial(1, 0.991) if mnationfr==. & mbirthfr==1
tab mbirthfr mnationfr, missing


//7. IMPUTATIONS MOTHER'FAMILY -> OTHERS' INFO

//7.1. GRANDMOTHER MOTHER
tab m_gmbirthfr m_gmnationfr, missing
replace m_gmnationfr = 0 if mnationfr==0 & m_gmnationfr==.
replace m_gmnationfr = 1 if mnationfr==1 & m_gfnationfr==0 & m_gmnationfr==.
tab m_gmbirthfr m_gmnationfr, missing

//7.2. GRANDFATHER MOTHER
tab m_gfbirthfr m_gfnationfr, missing
replace m_gfnationfr = 0 if mnationfr==0 & m_gfnationfr==.
replace m_gfnationfr = 1 if mnationfr==1 & m_gmnationfr==0 & m_gfnationfr==.
tab m_gfbirthfr m_gfnationfr, missing


//8. IMPUTATIONS MOTHER'FAMILY -> BIRTH PLACE AND NATIONALITY

//8.1. GRANDMOTHER MOTHER
tab m_gmbirthfr m_gmnationfr
replace m_gmbirthfr = rbinomial(1, 0.952) if m_gmnationfr==1 & m_gmbirthfr==.
replace m_gmbirthfr = rbinomial(1, 0.034) if m_gmnationfr==0 & m_gmbirthfr==.
replace m_gmnationfr = rbinomial(1, 0.992) if m_gmnationfr==. & m_gmbirthfr==1
replace m_gmnationfr = rbinomial(1, 0.181) if m_gmnationfr==. & m_gmbirthfr==0
tab m_gmbirthfr m_gmnationfr, missing


//8.2. GRANDFATHER MOTHER
tab m_gfbirthfr m_gfnationfr
replace m_gfbirthfr = rbinomial(1, 0.945) if m_gfnationfr==1 & m_gfbirthfr==.
replace m_gfbirthfr = rbinomial(1, 0.039) if m_gfnationfr==0 & m_gfbirthfr==.
replace m_gfnationfr = rbinomial(1, 0.99) if m_gfnationfr==. & m_gfbirthfr==1
replace m_gfnationfr = rbinomial(1, 0.191) if m_gfnationfr==. & m_gfbirthfr==0
tab m_gfbirthfr m_gfnationfr, missing




*********************** MIGRATION VARIABLES MOTHER *****************************

//1. Grandparents mother
g m_gmimm = 1 if m_gmbirthfr==0 & m_gmnationfr==0
replace m_gmimm = 0 if m_gmimm==. & !inlist(m_gmbirthfr,.) & !inlist(m_gmnationfr,.)
label var m_gmimm "Grandmother born abroad with foreign nationality (regardless of current residence place), mother"
label define m_gmimm 1 "yes"
tab m_gmimm if months2==1, missing

g m_gfimm = 1 if m_gfbirthfr==0 & m_gfnationfr==0
replace m_gfimm = 0 if m_gfimm==. & !inlist(m_gfbirthfr,.) & !inlist(m_gfnationfr,.)
label var m_gfimm "Grandfather born abroad with foreign nationality (regardless of current residence place), mother"
label define m_gfimm 1 "yes"
tab m_gfimm if months2==1, missing

**EXTRA IMPUTATION
tab m_gmimm m_gfimm if months==1, missing //WE CAN IMPUTE MORE
tab m_gmimm m_gfimm, missing
replace m_gmimm = rbinomial(1, 0.04) if m_gfimm==0 & m_gmimm==.
replace m_gfimm = rbinomial(1, 0.05) if m_gmimm==0 & m_gfimm==.
tab m_gmimm m_gfimm if months==1, missing


//2. Mother
g mimm = 1 if mbirthfr==0 & mnationfr==0 //immigrant mother
replace mimm = 2 if m_gmimm==1 & m_gfimm==1 & mbirthfr==1 & missing(mimm) //mother descendant of two immigrants

g m_oneimm = 1 if m_gmimm==1 & m_gfimm==0
replace m_oneimm = 1 if m_gmimm==1 & m_gfimm==.
replace m_oneimm = 1 if m_gmimm==0 & m_gfimm==1
replace m_oneimm = 1 if m_gmimm==. & m_gfimm==1

replace mimm = 3 if m_oneimm==1 & mbirthfr==1 & missing(mimm) //mother descendant of at least one immigrant
replace mimm = 4 if !inlist(mbirthfr,.) & !inlist(mnationfr,.)& missing(mimm) //rest of population

label var mimm "mother immigrant history"
label define mimm 1 "immigrant mother" 2 "mother descendant of two immigrants" 3 "mother descendant of at least one immigrant" 4 "rest of population"
label values mimm mimm
tab mimm, missing
tab mimm if months2==1, missing
tab mimm if months2==1 & !missing(m_gmimm), missing //we just keep if now missings from grandparents


***
//9. WHAT ABOUT ALL MISSING OBSERVATIONS??
//MOTHER
tab mbirthfr mnationfr, missing
tab mbirthfr mnationfr if months2==1, missing

//GRANDMOTHER, FATHER
tab m_gmbirthfr m_gmnationfr, missing
tab m_gmbirthfr m_gmnationfr if months2==1, missing

*afther all imputations, when we keep just women in 2 months survey,
*we have 53 missing values 
*keep if m_gmbirthfr==. & m_gmnationfr==. & months2==1


//GRANDFATHER, FATHER
tab m_gfbirthfr m_gfnationfr, missing
tab m_gfbirthfr m_gfnationfr if months2==1, missing

*afther all imputations, when we keep just women in 2 months survey,
*we have 219 missing values 
***




///////////////////////////////////////////////////////////////////////////////

**** EXTRA
**** MULTIVARIATE MULTIPLE IMPUTATION_ CHEINED
mi register imputed  f_gmbirthfr f_gmnationfr f_gfbirthfr f_gfnationfr // we indicate the set of variables to be imputed
mi register regular fbirthfr fnationfr fleng // variables to use in imputation -> must be full variables. without missings
mi describe

keep if months2==1 & f_unknown==0
mi impute chained (logit) f_gmbirthfr f_gmnationfr f_gfbirthfr f_gfnationfr = i.fbirthfr i.fnationfr i.fleng, add(10) force //10 number of imputations


*=============================== THE END =====================================