*********************************************************************************
*************** ZA 2010 conversion to MTUS X ************************************
*********************************************************************************


use "ZA2010_activities.dta" , clear

*makes data view easier to look at
format Activity_code %8.0g
format Sametime %8.0g
format Location1 %8.0g
format Location2 %8.0g
format Q22HighestSchool %8.0g
format Q23MaritalStatus %8.0g

order hldid persid Timeslot slot
sort hldid persid Timeslot slot

*flaging slots when new activity starts
bysort hldid persid : gen epact = Activity_code if (Activity_code[_n] !=  Activity_code[_n-1])

*this code creates a sequence variable for episode number 
*--------------
preserve

drop if epact == .
bysort hldid persid: gen ep_n = _n	/*create episode sequence number*/
sort hldid persid Timeslot Act
keep hldid persid Timeslot Act ep_n 
save "ep_n.dta"

restore
*--------------

*merge new episode sequence number into working file
merge 1:1 hldid persid Timeslot Act using "ep_n.dta"

drop _merge 

*delete file just merged
rm "ep_n.dta"

*fill in blanks for new episode sequence number
replace ep_n = ep_n[_n-1] if ep_n[_n] == .
*fill in blanks for new activity variable (you can actually delete this and use the old one)
replace epact = epact[_n-1] if epact[_n] == .

drop slot dur total_time

replace time = 7.5 if Timeper == 7.5 & time == 7			//somehow the time variable was 7 and should be 7.5

*this created new episode file with unique case for each new episode
*note rule (first) is applied to all variables except 'time' which is summed (this is the episode duration variable)
collapse (first) Timeslot Act Q116Gender Q117Age Q118Population Q22HighestSchool ///
Q23MaritalStatus Activity_code Sametime Location1 Location2 Fulltime ///
Activ Timeper Weight sec unid epact (sum) time , by(hldid persid ep_n)

*check all is well with the world by computing total time variable
bysort hldid persid: egen tottime = sum(time)
ta tottime



* epnum *

gen epnum = ep_n


* clockst *

gen clockst = .
replace clockst = 4.0 if Timeslot==1
replace clockst = 4.3 if Timeslot==2
replace clockst = 5.0 if Timeslot==3
replace clockst = 5.3 if Timeslot==4
replace clockst = 6.0 if Timeslot==5
replace clockst = 6.3 if Timeslot==6
replace clockst = 7.0 if Timeslot==7
replace clockst = 7.3 if Timeslot==8  
replace clockst = 8.0 if Timeslot==9
replace clockst = 8.3 if Timeslot==10	  
replace clockst = 9.0 if Timeslot==11
replace clockst = 9.3 if Timeslot==12
replace clockst = 10.0 if Timeslot==13 
replace clockst = 10.3 if Timeslot==14
replace clockst = 11.0 if Timeslot==15
replace clockst = 11.3 if Timeslot==16
replace clockst = 12.0 if Timeslot==17
replace clockst = 12.3 if Timeslot==18  
replace clockst = 13.0 if Timeslot==19
replace clockst = 13.3 if Timeslot==20
replace clockst = 14.0 if Timeslot==21   
replace clockst = 14.3 if Timeslot==22
replace clockst = 15.0 if Timeslot==23
replace clockst = 15.3 if Timeslot==24
replace clockst = 16.0 if Timeslot==25   
replace clockst = 16.3 if Timeslot==26
replace clockst = 17.0 if Timeslot==27
replace clockst = 17.3 if Timeslot==28
replace clockst = 18.0 if Timeslot==29
replace clockst = 18.3 if Timeslot==30
replace clockst = 19.0 if Timeslot==31
replace clockst = 19.3 if Timeslot==32
replace clockst = 20.0 if Timeslot==33  
replace clockst = 20.3 if Timeslot==34
replace clockst = 21.0 if Timeslot==35
replace clockst = 21.3 if Timeslot==36   
replace clockst = 22.0 if Timeslot==37
replace clockst = 22.3 if Timeslot==38
replace clockst = 23.0 if Timeslot==39
replace clockst = 23.3 if Timeslot==40    
replace clockst = 0.0 if Timeslot==41
replace clockst = 0.3 if Timeslot==42 
replace clockst = 1.0 if Timeslot==43
replace clockst = 1.3 if Timeslot==44
replace clockst = 2.0 if Timeslot==45   
replace clockst = 2.3 if Timeslot==46
replace clockst = 3.0 if Timeslot==47
replace clockst = 3.3 if Timeslot==48
    
    

* start & end *

gen start=.
replace start = 0 if Timeslot==1

gen end =.
replace end = start+ time

replace start = end[_n-1] if hldid== hldid[_n-1] & persid ==persid[_n-1]
replace end = start+ time if end==.


bys hldid persid: egen max_end = max(end)
fre max_end


* activity codes *

* main/ sec*

gen main=Activity_code
recode main (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)

recode main (88 488=67) (188=63) (288 388=68) (588=66) (788=64) (888 = 68)

* sec*

replace sec = 69 if acta ==69   
recode sec (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)

recode sec (88 488=67) (188=63) (288 388=68) (588=66) (788=64) (888 = 68)
recode sec (777=7)

recode sec (.=-8)

* 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)

recode av (88 488=17)  (188=5) (288 388 588 788 888 = 17)

*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) 




************
* location *
************


* eloc *

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

* mtrav *

gen mtrav=Location2
recode mtrav (4 5 =1) (6 7 =2) (8=4) (9=5)

* ict *

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


* co-presence *

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


* id *

gen id=1

* diary *
gen diary=1


* inout*

gen inout= Location2

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

* animal *

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



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

* country *
gen country = "ZA"

* survey *

gen survey = 2000

*swave*
gen swave=0

* msamp*
gen msamp=0

* sex *

gen sex = Q116Gender


* age *

gen age =Q117Age



* 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


*****************************************************************
* generating & merging in background variables from other files *
*****************************************************************

* from person.dta *

destring persid, replace

merge m:1 hldid persid using tus-2010-person-v1-20140409.dta, ///
keepusing(Q22HighestSchool Q23MaritalStatus ///
Q118Population Q24Spouse Q23MaritalStatus Q26Child18Alive ///
Q31PdWrk Q41Occupation Q46TotIncome Occup Province ///
Education_Status Weight civstat cohab nchild partid agekidx ///
agekid2 day cday month year diary ocombwt workhrs empstat emp ///
unemp edcat sector region empinclm occupo educa ethnic ///
migrantd migrantm migrantf isco1 retired student)


merge m:1 hldid persid using tus-2010-person-v1-20140409.dta, ///
keepusing(civsum)

* day *

gen day =Q52DayDiary
recode day (7=1) (1=2) (2=3) (3=4) (4=5) (5=6) (6=7)

* cday *

tostring Q51DateDiary, gen(date_str)
gen str cday = substr(date_str, 1, strlen(date_str) - 6)
destring cday, replace

* month *
gen str month = substr(date_str, -6,2)
destring month, replace

* year *

gen year=2010

* diary *

gen diary=1

* nowght *

* parntid1 parntid2 *

gen parntid1=-8
gen parntid2=-8

*partid *

gen partid_n = Q25Spouse

gen partid = -8

*hhtype *

bys hldid: egen civsum= total(civstat) 

gen hhtype=.
replace hhtype = 1 if hhldsize==1
replace hhtype=2 if hhldsize==2 & civsum>=1
replace hhtype=3 if hhldsize>=3 & civsum>=1
replace hhtype=4 if hhldsize>=3 & civsum==0

*famstat *

gen famstat = 0 if (age >=18 & age <=39) & nchild ==0
replace famstat =1 if age >=18 & agekidx==2 // under 7
replace famstat =2 if age >-18 & agekidx==3 // under 18, over 7
replace famstat =3 if age >=40 & nchild==0
replace famstat =5 if age <18 & famstat ==.

replace famstat = 

*hhldsize *

gen hhldsize =Q122NumberEligible


* nchild *

gen nchild =Q27Child18HH
* agekidx *

gen agekidx = .
replace agekidx = 2 if Q29Child06HH>=1 & Q29Child06HH!=88
replace agekidx = 3 if agekidx!=2 & Q27Child18HH<=9 & Q29Child06HH!=0


* agekid2 *

gen agekid2=-8

*income *
// total hh income grouped //

gen income =.
replace income=-7 if incorig==0 | incorig==1
replace income = 1 if incorig==2 | incorig==3 | incorig==4 | incorig==5
replace income = 2 if incorig==6 | incorig==7
replace income = 3 if incorig>7 & incorig <13
replace income = -8 if incorig>=13

*ownhome *
gen ownhome=-8

*urban *
gen urban=-8

*computer*

gen computer=1 if Q12Computer==1
replace computer=0 if Q12Computer==2

*vehicle *

gen vehicle=1 if Q12Car==1
recode vehicle = 0 if Q12Car==2


*singpar *
gen singpar=-8

*relrefp*
gen relrefp=-8

*civstat*

gen civstat=.
replace civstat =1 if Q23MaritalStatus==1 | Q23MaritalStatus==2
replace civstat = 2 if Q23MaritalStatus>=3

* cohab *

gen cohab =.
replace cohab= -7  if civstat==2
replace cohab=1 if Q23MaritalStatus==2
replace cohab=0 if Q23MaritalStatus==1

*citizen*
gen citizen =-8

* whereborn*
gen whereborn=-8

* empstat*
gen empstat = 3 if Status==1
replace empstat = 4 if Status==2 | Status==3
replace empstat=-8 if empstat==.

* emp*
gen emp=1 if Status==1
replace emp=0 if emp==.

* unemp* 
gen unemp=1 if Status ==2
replace unemp=0 if unemp==.

*student* 

gen student=-8

*retired*

gen retired= 1 if Q45SourceIncome==3 | Q45SourceIncome==6
replace retired=0 if retired !=1

 
*empsp*

gen empsp=-8

gen empsp_n=.

gen empsp_help = 1 if civstat ==1 //flagging partners//
replace empsp_help = empstat if empsp_help ==1

*on consecutive partners (one after another*

gen civstat_help=1 if partid>=1

bys hldid: replace empsp = empsp_help[_n-1] if civstat_help !=. //uwaga: rekoduje tez w drugim HH jesli 2 HH obok siebie!//
bys hldid: replace empsp = empsp_help[_n+1] if civstat_help !=. & empsp_help[_n+1] !=. & hldid == hldid[_n+1] & empsp ==.

* if partners on different numbers within HH (split by one)*
bys hldid: replace empsp = empsp_help[_n-2] if empsp_help[_n-1]==. & civstat_help !=. & empsp_help[_n-2] !=. & empsp ==.
bys hldid: replace empsp = empsp_help[_n+2] if empsp_help[_n+1]==. & civstat_help !=. & empsp_help[_n+2] !=. & empsp ==.

* other cases that cannot be captured by macro *


*workhrs *
gen workhrs = Q43Hours

*isco1 *
gen isco1=Occup
recode isco1 (10=9)

*sector*

gen sector=-8
* edcat* 

gen edcat = Education_Status
recode edcat (1 2 3 4=1) (5=2) (6 7=3)



*rushed *
gen rushed=-8
*health *
gen health=-8
*carer *
gen carer=-8
*disab*
gen disab=-8
* ocombwt *

gen ocombwt = Weight

*propwt*
egen count=count(badcase)
egen countp_bis=count(badcase) if badcase==0
egen countp=min(countp_bis)
 
sort sex age day
egen daywt = group(sex age day) if badcase==0
egen ngroupsd=max(daywt) if badcase==0
by sex age day: egen daycount2=count(daywt) if badcase==0
 
sort sex age
egen weekwt=group(sex age) if badcase==0
by sex age: egen wkcount2=count(weekwt) if badcase==0
 
gen propwt=((wkcount2/7)/daycount2) if badcase==0
replace propwt=0 if badcase!=0
replace propwt=propwt*(count/countp)
 
egen mean_weight=mean(propwt)
replace propwt=propwt/mean_weight


************************
*  from household.dta  *
************************


merge m:1 hldid using tus-2010-household-v1-20140409.dta, keepusing(computer ///
vehicle incorig region hhldsize)



*************************
* preparing final files *
*************************

* drop total time under 1440 *

drop if max_end !=1440

********
* 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


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

* from person.dta *


* region *

gen region = Province
label val region Province 

* incorig *

gen incorig = Q113Income
label val incorig Q113Income

 
* empinclm *

gen empinclm = Q46TotIncome
label val empinclm TotIncome



* occupo *

gen occupo = Occup
label val occupo Occup 



* educa *

gen educa = Education_Status
label val educa Education_Status


* ethnic *

gen ethnic = Q118Population

label define ethnic ///
1	"African/Black" ///
	2 "Coloured" ///
	3 "Indian/Asian" ///
	4 "White" ///

label val ethnic ethnic


* migrantd *

gen migrantd = -9
* migrantm *

gen migrantm=-9

* migrantf *
gen migrantf=-9


* data prep *

keep if epnum==1


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