**************************************************************************************************
************************** ZA 2000 MTUSX CONVERSION **********************************************


* main/ sec*

gen main=acta
recode main ( -8=69) (-7=1) (-6=3) (10=2) (30 90=4) 
replace main = 5 if acta ==20 & (loca ==3 | loca ==4)
recode main (41 42 43 48 = 25) (50=55) (60 820=34) (80=67) 
recode main (111 = 7)
recode main (115 130 190 210/260 290 310 320 330 340 350 360 370=777)
recode main (112=8)
replace main = 8 if main == 777 & loca==1
replace main = 7 if main == 777 & (loca ==3 | loca==4)
recode main (113 = 23)
recode main (114=  13)
recode main (140=12)
recode main (150=14)     
recode main (180=63)
recode main (280=11)
recode main (370 390=9)
recode main (410=18)
recode main (420=20)
recode main (430 = 21)
recode main (440=24)
recode main (441 448=26)
recode main (450 460 490=22)
recode main (470=27)
recode main (480=67)
recode main (511 512 590= 28)
recode main (491=22)
recode main (521 522 =29)
recode main (531 532 561 562=31)
recode main (540 550 673=32) (580 =66) (610 615 620 630 650 660 671 672 674 690=33)
recode main (680=65)
recode main (710=15) (720=16) (730 740 790 =17) (780=64)
recode main (810 =40) (832=48) (831 833=49) (840=52) (870=35) (850=42) (860 890=50)
recode main (880=68) (910 = 56) (920 990=59) (930=58) (940=61) (950=38) (980=68)
replace main =7 if main ==777 & loca !=1
recode main (380=11)


gen sec = actb
replace sec = 69 if acta ==69   
recode sec (-8=69) (-7=1) (-6=3) (10=2) (30 90=4) 
replace sec = 5 if acta ==20 & (loca ==3 | loca ==4)
recode sec (41 42 43 48 = 25) (50=55) (60 820=34) (80=67) 
recode sec (111 = 7)
recode sec (115 130 190 210/260 290 310 320 330 340 350 360 370=777)
recode sec (112=8)
replace sec = 8 if main == 777 & loca==1
replace sec = 7 if main == 777 & (loca ==3 | loca==4)
recode sec (113 = 23)
recode sec (114=  13)
recode sec (140=12)
recode sec (150=14)     
recode sec(180=63)
recode sec (280=11)
recode sec(370 390=9)
recode sec(410=18)
recode sec (420=20)
recode sec (430 = 21)
recode sec (440=24)
recode sec (441 448=26)
recode sec (450 460 490=22)
recode sec (470=27)
recode sec (480=67)
recode sec (511 512 590= 28)
recode sec (491=22)
recode sec (521 522 =29)
recode sec (531 532 561 562=31)
recode sec (540 550 673=32) (580 =66) (610 615 620 630 650 660 671 672 674 690=33)
recode sec (680=65)
recode sec (710=15) (720=16) (730 740 790 =17) (780=64)
recode sec(810 =40) (832=48) (831 833=49) (840=52) (870=35) (850=42) (860 890=50)
recode sec (880=68) (910 = 56) (920 990=59) (930=58) (940=61) (950=38) (980=68)
replace sec = 8 if sec ==777 & loca ==1
recode sec (380=11)


* av *

gen av=acta
recode av (111 113 115 130 140 190 310 320 330 340 350 360 370 390=1)
recode av (112 150=2)
//av3=-8//
recode av (710 730 740=4)
recode av (180 380 780=5)
recode av (410=6)
recode av (420 430=7)
recode av (236 240 250 450 460 470 490 491 540 673=8)
recode av (210 220 230 290=9)
recode av (260 440 441 448=10)
recode av (511 512 521 522 561 562 590 671 672=11)
recode av (80 280 480 531 532 550 580=12)
recode av (30=13)
recode av (41 42 43 48 90=14)
recode av (20=15)
recode av (10=16)
recode av (680 880 980=17)
recode av (870 950=18)
recode av (850=19)
//av 320 21 24 26 27 28 32 37 38 39= -8//

recode av (60 820=22)
recode av (114 610 615 620 630 650 660 674 690=23)
recode av (810 890=25)
recode av (831 832 833=29)
recode av (930=30)
recode av (920 990=31)
recode av (720 790 940=33)
recode av (910=34)
recode av (50=36)
recode av (840 860=40)
recode av (-7 -6 -8=41)



* hldid *

rename uqnr hldid

* id *

gen id=1

* diary *
gen diary=1


* inout*

gen inout= LOCB

recode inout (1=1) (2 3 8=2) (4/7 9=3)


*TUTUTUTUU*

* eloc *

gen eloc = loca
recode eloc (3 4 =3) (5=4) (7=8) (6 8=9)

* mtrav *

gen mtrav=LOCB
recode mtrav (4 5 =1) (6 7 =2) (8=4) (9=5)

* ict *

gen ict = 1 if acta ==940 | actb==940
recode ict (.=0)


* co-presence *

gen alone=-8
gen child=-8
gen sppart = -8
gen oad = -8


*core25*
gen core25 = main
replace core25 = 1 if main ==2 | main==3
replace core25 = 2 if main ==5 | main==6
replace core25 = 3 if main ==1 | main==4 
recode core25(7 8 9 10 11 12 13 14 =4) (15 16 17 =5) ///
(18 19=6) (20 21 23 = 7) (22=8) (24 25 26=9) (46=10) (27 47=11) (32=12) (28 31=13) (29 30=14) ///
(34=15) (33=16) (63 64=17) (62 65 66 67 68=18) (42 43 44=19) (57 58 59=20) (56=21) ///
(60 61=22) (35 36 37 38 39 40 41 45=23) (48 49 50 51 52 53 54 55=24) (69=25) 


***********************************
*      BACKGROUND VARIABLES       *
***********************************

* country *
gen country = "ZA"

* survey *

gen survey = 2000

*swave*
gen swave=0

* msamp*
gen msamp=0



* persid *

rename person persid



* TUTUTUTUTTTTTTUUUUU*

* region *
// from ZA2000_hhld.dta//
gen region= PROVINCE

* merging in background variables *

merge m:1 hldid persid using SouthAfrica_2000.dta, keepusing (month ///
year hhldsize nchild agekidx income urban sex age civstat empstat workhrs ///
empinclm edcat propwt sppart)



rename merge merge_1


* from ZA2000_member.dta *

merge m:1 hldid persid using ZA2000_member.dta, keepusing(DAYDIARY OCCUPATI TOTINCOM ///
ECSTAT PROV PERWGT INDUS OCCUP ALLSECT NOTLOOKI)


* day *

gen day = DAYDIAR
recode day (0=-8) (1=2) (2=3) (3=4) (4=5) (5=6) (6=7)(7=1)

*cday *

gen cday=-8


* time *

gen time = end - start


* badcase *

bys hldid persid: egen max_epi = max(epnum)

gen badcase =.
replace badcase = 4 if max_epi <=7
replace badcase = 1 if sum_mis >90
replace badcase = 0 if badcase==.

gen mis=time if main==.
bys hldid persid: egen sum_mis = sum(mis)


*nowght*

gen nowght = 1 if badcase !=0


* reference pers *

gen parntid1 =-8
gen parntid2 =-8
gen partid =-8
gen relrefp = -8


* animal *

gen animal = 1 if main==27 | sec==27
replace animal = 0 if animal !=1

*agekid2*

gen agekid2 = -9

*cohab *

gen cohab=-9

*singpar*

gen singpar = -9

* empsp *

gen empsp = -9


* student *
gen student =1 if NOTLOOKI==1
replace student =0 if NOTLOOKI !=1

* retired *

gen retired =1 if NOTLOOKI ==2
replace retired =0 if NOTLOOKI !=2


*disab *

gen disab =1 if NOTLOOKI==3
replace disab =0 if NOTLOOKI!=3

* ocombwt *

gen ocombwt = PERWGT

* carer *

gen carer =-9

* isco1 *

gen isco1= OCCUP
recode isco1 (10=9)

//inforamtion - isco 9 includes also domestic workers //

* emp *

gen emp= 1 if ECSTAT ==1
replace emp = 0 if ECSTAT !=1

* unemp *

gen unemp = 1 if ECSTAT ==2
replace unemp =0 if ECSTAT !=2

* sector *

gen sector =-9


* hhtype *

gen hhtype =1 if hhldsize ==1
replace hhtype =2 if civstat==1 & hhldsize ==2
replace hhtype =3 if civstat==1 & hhldsize > 2
replace hhtype =4 if hhtype ==.

*famstat *

gen famstat = 0 if (age >=18 & age <=39) & nchild ==0
replace famstat =1 if age >=18 & agekidx==1 // under 5
replace famstat =2 if age >-18 & (agekidx ==2 | agekidx==3) // 5-17, none under 5
replace famstat =3 if age >=40 & nchild==0
replace famstat =5 if age <=18 & famstat ==.


* merge in from household file *

merge m:1 hldid using ZA2000_hhld.dta, keepusing(q12car PROVINCE)

*ownhome *

gen ownhome =-9

*computer *

gen computer=-9

*vehicle *

gen vehicle = q12car
*citizen *

gen citizen =-9

*whereborn *
gen whereborn=-9

*rushed *

gen rushed =-9

*health *

gen health =-9

* region *

gen region = PROVINCE

label define region ///
           1 "Western Cape" ///
           2 "Eastern Cape" /// 
           3 "Northern Cape" /// 
           4 "Free State" ///
           5 "KwaZulu-Natal" ///
           6 "North West" ///
           7 "Gauteng" ///
           8 "Mpumalanga" ///
           9 "Northern Province"

label val region region

******************
* final cleaning *
******************
drop if merge1==2 | merge2==2 | merge3 ==2 | merge4==2




********
* MAIN *
********

keep country survey swave msamp hldid persid id day cday month year diary nowght ///
time clockst start end epnum main core25 sec av inout eloc ict mtrav alone child sppart oad ///
animal parntid1 parntid2 partid hhtype hhldsize nchild agekidx agekid2 income ownhome urban computer ///
vehicle sex age famstat singpar relrefp civstat cohab citizen whereborn empstat emp unemp student retired empsp ///
workhrs isco1 sector edcat rushed health carer disab ocombwt propwt

order country survey swave msamp hldid persid id day cday month year diary nowght ///
time clockst start end epnum main core25 sec av inout eloc ict mtrav alone child sppart oad ///
animal parntid1 parntid2 partid hhtype hhldsize nchild agekidx agekid2 income ownhome urban computer ///
vehicle sex age famstat singpar relrefp civstat cohab citizen whereborn empstat emp unemp student retired empsp ///
workhrs isco1 sector edcat rushed health carer disab ocombwt propwt




* FROM MASTERFILE *

**********
* SURVEY *
**********

merge m:1 hldid using ZA2000_hhld.dta, keepusing(q113inco)

* incorig *

gen incorig =q113inco


 merge m:1 hldid persid using ZA2000_member.dta, keepusing(RACE HIGHESTS STUDY)
 
* empinclm *

gen empinclm = TOTINCOM

label val empinclm TOTINCOM


* occupo *

gen occupo = OCCUP

label val occupo OCCUP

* educa *

gen educa = HIGHESTS


* ethnic *

gen ethnic = RACE

label ethnic RACE

* migrantd *

gen migrantd = -9
* migrantm *

gen migrantm=-9

* migrantf *
gen migrantf=-9

keep country survey swave msamp hldid persid incorig region empinclm occupo ///
educa ethnic migrantd migrantm migrantf

sort country survey swave msamp hldid persid incorig region empinclm occupo ///
educa ethnic migrantd migrantm migrantf