// Program:	kr2009_mtusx.do
// Task:	Korean MTUS-X main and survey files
// Project:	MTUS-X file
// Created:	07/11/2019



//	Program setup
******************************************************************************

set more off
set linesize 80
set varabbrev off
clear all


// Variables
******************************************************************************

use kr2009_sequence, clear

sort col2 col3 //sort by hhid and persid

//Household variables
gen country = "KR"
gen survey=2009
gen swave=0 //longitudinal
gen msamp=0 //multiple samples
gen hldid=col2 //hhid
gen persid=col3 //personal id
seq a, f(1) t(2) by(hldid persid)//diary identifier 
rename a id
gen day=col5
gen cday=-9
gen month=9
gen year=2009

gen diary=0 //diary order
replace diary=1 if id==1
replace diary=2 if id==2
sort hldid

gen animal = -9

by hldid: gen chcount=1 if h_age<18 //age of younget child
by hldid: egen nchild=count(chcount)
by hldid: replace nchild=0 if nchild==.

gen nchild2=nchild/2
drop nchild
rename nchild2 nchild //number of child aged under 18

egen hld_tag=tag (hldid persid) //hhldsize: number of people in household
egen hhldsize = sum(hld_tag), by(hldid)

sort hldid

by hldid: gen chage =h_age if h_age < 18 //Identify those whose ages are below 18
sort hldid chage
by hldid: replace chage = chage[1]

by hldid: replace chage =h_age if chage == . & (rela_h ==3|rela_h ==4) //Identify those who are children of the reference person
by hldid: gen inlaw_age=1 if chage>17 & rela_h ==4 //Mark cases with in-law can be the youngest child

sort hldid chage
by hldid: replace chage = chage[1]

sort hldid rela_h 
by hldid: gen parn = 1 if rela_h==6
sort hldid parn
by hldid: replace parn = parn[1]
sort hldid rela_h 
by hldid: replace chage =h_age if chage == . & rela_h == 1 & parn == 1

by hldid: egen young = min(chage)

replace young = -7 if young == .
gen agekid = 1 if (young>=0 & young<=4)
replace agekid = 2 if (young>=5 & young<=12)
replace agekid = 3 if (young>=13 & young<=17)
replace agekid = 4 if young >= 18
replace agekid=-7 if young==-7

rename agekid agekidx

sort hldid
egen pregnant=tag(hldid) if agekidx==.
tab pregnant

list hldid if pregnant==1 //three households with pregnant women
replace agekidx=1 if agekidx==.

gen agekid2 = young //age of youngest child
recode agekid2 (-1=0)

gen income=-9

gen ownhome=-5
replace ownhome=1 if col8==1
replace ownhome=2 if col8==2|col8==3
replace ownhome=3 if col8==4


//household income
recode empinclm (1=0)(2=25)(3=75)(4=125)(5=175)(6=225)(7=275)(8=325)(9=375) ///
	(10=425)(11=500), into(make_income)

egen med_income=sum(make_income) if id==1, by(hldid)
xtile quart = med_income, nq(4)

gen income=-5
replace income=1 if quart==1
replace income=2 if quart==2|quart==3
replace income=3 if quart==4

sort hldid persid 
save kr2009_hh.dta, replace


//Individual variables
gen civstat=-5 
replace civstat=1 if col14==2
replace civstat=2 if col14==1|col14==3|col14==4
replace civstat=-8 if col14==.

gen sex=col11 

gen age=col12
recode age (90/max=90) //cap is 90

gen relrefp=-5
replace relrefp=1 if rela_h==1
replace relrefp=2 if rela_h==2
replace relrefp=3 if rela_h==3 |rela_h==4 
replace relrefp=4 if rela_h==6 
replace relrefp=5 if rela_h==8
replace relrefp=9 if rela_h==5 | rela_h==7 
replace relrefp=10 if rela_h==9
replace relrefp=-8 if rela_h==.
 
*keep if id==1 //then run through for id = 2 

//partid
//Identify the spouse/partner of the reference person
by hldid: gen spid = persid if relrefp == 2
sort hldid spid
by hldid: replace spid = spid[1] if spid == .
by hldid: gen refid = persid if relrefp == 1
sort hldid refid
by hldid: replace refid = refid[1] if refid == .
by hldid: gen partid = refid if relrefp == 2
by hldid: replace partid = spid if relrefp == 1

//Identify the spouse/partner of non-reference person
by hldid: gen inlaw_m = persid if (rela_h== 4 & sex==1 & civstat!=2)
by hldid: gen inlaw_f = persid if (rela_h ==4 & sex ==2 & civstat!=2) 

by hldid: gen ninlaw=1 if (inlaw_m!=.|inlaw_f!=.) 
by hldid: egen total_inlaw=count(ninlaw)

by hldid: replace total_inlaw=0 if total_inlaw==.
count if total_inlaw==2 & ninlaw==1 

sort hldid inlaw_m inlaw_f
by hldid: replace inlaw_m=inlaw_m[1] if inlaw_m==.& ninlaw==1 & total_inlaw==2

sort hldid inlaw_m inlaw_f
by hldid: replace inlaw_f=inlaw_f[1] if inlaw_f==. &ninlaw==1 & total_inlaw==2

sum age if total_inlaw==2 & ninlaw==1, detail

sort hldid inlaw_m inlaw_f
by hldid: replace partid = inlaw_m if ninlaw==1 & total_inlaw==2 & sex==2 & partid==. & civstat!=2

sort hldid inlaw_m inlaw_f
by hldid: replace partid = inlaw_f if ninlaw==1 & total_inlaw==2 & sex==1 & partid==. & civstat!=2

replace civstat=1 if partid!=. & civstat!=2

sort hldid persid
by hldid: gen non=1 if h_suv!=1 & civstat!=-8

sort hldid non 
by hldid: replace non=non[1] if partid!=.

replace partid = -8 if non==1

tab partid non if partid==-8

replace partid = -7 if partid == .

replace partid = -8 if partid==-7 & civstat==1

//parntid1 & panrtid2
sort hldid 
by hldid: gen parntid = persid if relrefp == 4

sort hldid relrefp persid
by hldid relrefp: gen parnt_no = _n if relrefp == 4

by hldid: gen parntid1 = parntid if parnt_no == 1
by hldid: gen parntid2 = parntid if parnt_no == 2
sort hldid parntid1 
by hldid: replace parntid1 = parntid1[1] if parntid1 == .
sort hldid parntid2
by hldid: replace parntid2 = parntid2[1] if parntid2 == .
replace parntid1 = . if relrefp ~= 1
replace parntid2 = . if relrefp ~= 1

sort hldid 
by hldid: gen child = 1 if relrefp == 3

sort hldid child
by hldid: replace child = child[1] if child == .
by hldid: gen parent1 = persid if relrefp == 1 & child == 1
sort hldid parent1
by hldid: replace parent1 = parent1[1] if parent1 == . & child == 1
by hldid: gen parent2 = persid if relrefp == 2 & child == 1
sort hldid parent2
by hldid: replace parent2 = parent2[1] if parent2 == . & child == 1
replace parent1 = . if relrefp ~= 3
replace parent2 = . if relrefp ~= 3

replace parntid1 = parent1 if parntid1 == . & parent1 ~= .
replace parntid2 = parent2 if parntid2 == . & parent2 ~= .

sort hldid col10 
by hldid: gen grandptid = persid if col10 ==7 

order hldid col10 persid grandptid parntid1 parntid2 

sort hldid col10 persid 
by hldid col10: gen grandpt_no = _n if col10 ==7

order hldid col10 persid grandptid grandpt_no parntid1 parntid2 

sort hldid grandptid  
by hldid: replace grandptid = grandptid[1] if grandptid==.

replace grandptid = . if col10 ~=6
replace parntid1=grandptid if parntid1==. & col10==6
drop grandptid grandpt_no

sort hldid 
by hldid: gen sibling=1 if relrefp==5

sort hldid sibling
by hldid: replace sibling =sibling[1] if sibling==.

by hldid: gen npar=1 if (sibling==1 & relrefp==4)

order hldid persid relrefp npar sibling parntid1 parntid2 

sort hldid relrefp 
by hldid: replace parntid1= parntid1[1] if parntid1==. & relrefp==5
by hldid: replace parntid2= parntid2[1] if parntid2==. & relrefp==5
replace parntid1 = -7 if parntid1 == . 
replace parntid2 = -7 if parntid2 == . 

sort hldid

by hldid: gen grandchild=1 if col10==5

sort hldid grandchild
by hldid: replace grandchild=grandchild[1] if grandchild==.
by hldid: gen midpar=1 if (grandchild==1 & relrefp==3)

sort hldid col10 persid 
by hldid col10: gen midpar_no = _n if grandchild==1 & relrefp==3

by hldid: gen m_parntid = persid if grandchild==1 & relrefp == 3

by hldid: gen mparntid1=m_parntid if midpar_no==1
by hldid: gen mparntid2=m_parntid if midpar_no==2 

sort hldid mparntid1
by hldid: replace mparntid1=mparntid1[1] if mparntid1==.

sort hldid mparntid2
by hldid: replace mparntid2=mparntid2[1] if mparntid2==.

by hldid: replace parntid1=mparntid1 if parntid1==-7 & col10==5
by hldid: replace parntid2=mparntid2 if parntid2==-7 & col10==5

order hldid persid relrefp col10 grandchild midpar_no m_parntid mparntid1 mparntid2 parntid1 parntid2

replace parntid1 = -7 if parntid1 == . 
replace parntid2 = -7 if parntid2 == . 

list hldid if col10==5 & parntid1==-7


//household type
gen hhtype=-5

gen couple=1 if civstat==1

sort hldid
by hldid: egen hcouple=count(couple)

sort hldid

by hldid: replace hhtype=1 if hhldsize==1 //single person household
by hldid: replace hhtype=2 if hhldsize==2 & hcouple==2 //couple only

by hldid: replace hhtype=3 if hhldsize>2 & hcouple>1 //couple + others
by hldid: replace hhtype=3 if hhldsize==2 & hcouple==1
by hldid: replace hhtype=3 if hhldsize>2 & hcouple==1


replace civstat=1 if hldid==1052 & empsp==-8 & civstat==2 //correcting - dead spouse
replace hhtype =2 if hldid==1052 

by hldid: gen hhtype_m=1 if (hhldsize>=2 & hcouple==1)

list hldid if hhldsize>=2 & hcouple==1 

list hldid if h_suv!=1 & h_suv!=. & h_suv!=2

count if hhtype_m==1

egen tag=tag(hldid hhtype_m)
egen count2=total(tag)
tab count2

by hldid: replace hhtype=4 if hhldsize>=2 & hcouple==0 //other

tab hhtype civstat, missing

//famstat
gen famstat=-5
replace famstat = 0 if nchild == 0 & age >= 18 & age <= 39
replace famstat = 1 if famstat ==-5 & age>= 18 & agekidx == 1
replace famstat=2 if famstat==-5 & (age>=18 & agekidx==2| agekidx==3)
replace famstat=3 if famstat==-5 & nchild==0 & age>=40
replace famstat = 4 if age < 18 & famstat==-5 & ((parntid1!=-7)| (parntid2!=-7))
replace famstat = 5 if age < 18 & famstat==-5 & (parntid1==-7 & parntid2==-7)

//singpar
gen singpar=-5

sort hldid 
by hldid: gen parent=1 if relrefp == 4 //parents

sort hldid child
by hldid: gen child = 1 if relrefp == 3 
by hldid: replace child=child[1] if child==.

sort hldid parent
by hldid: replace parent=1 if relrefp ==1 & child==1  
by hldid: replace parent = 1 if relrefp == 2 & child == 1

by hldid : replace singpar=1 if parent==1 & (partid==-7)
by hldid : replace singpar=0 if parent==.| (parent==1 & partid!=-7)

tab singpar civstat if h_suv==1

replace singpar2=0 if 

gen singpar2=-5
replace singpar2=0 if col36!=1 & col36!=2
replace singpar2=1 if col36==1 | col36==2

gen ownhome=-5 //ownhome
replace ownhome=1 if col8==1
replace ownhome=2 if col8==2|col8==3
replace ownhome=3 if col8==4

gen urban=-5 //urban
replace urban=1 if col6==2
replace urban=2 if col6==1

gen computer=-9

gen vehicle=-9

gen cohab=-5 //cohabiting
replace cohab=0 if col14==2
replace cohab=1 if (col14==1|col14==3|col14==4) & (partid!=-7)
replace cohab=-7 if (col14==1|col14==3|col14==4) & (partid==-7)

gen citizen=1
gen whereborn = -9
gen health=-9
gen carer=-9
gen disab=-9

gen workhrs=-5
replace workhrs = col24 + col25 if col24>=0 & col25>=0 //main + second jobs
replace workhrs=col24 if (col24>=0 & col25<0)
replace workhrs=col25 if (col24<0 & col25>=0)
replace workhrs = -7 if col22==22 

gen empstat=-5
replace empstat=1 if col29==1 |((col28==2|col28==3|col28==4)& workhrs>=30)
replace empstat=2 if col29==2 |((col28==2|col28==3|col28==4)&(workhrs>0 & workhrs<30))
replace empstat=3 if (empstat==-5 & workhrs==0)
replace empstat=4 if paidwork==2

gen emp=-5
replace emp=1 if paidwork==1
replace emp=0 if paidwork==2

gen unemp=-9

gen student=-5 
replace student=1 if col16 ==2|col16==4
replace student=0 if col16 ==1 | col16==3| col16==5
replace student=0 if col16==.

gen retired=0
replace retired=1 if age>59 & (empstat!=1)

gen empsp=-5
sort hldid
by hldid: gen spid = empstat if relrefp == 2
sort hldid spid
by hldid: replace spid = spid[1] if spid == .
by hldid: gen refid = empstat if relrefp == 1
sort hldid refid
by hldid: replace refid = refid[1] if refid == .
by hldid: replace empsp = refid if relrefp == 2
by hldid: replace empsp = spid if relrefp == 1

by hldid: gen einlaw_m = empstat if (col10 == 4 & sex==1)
by hldid: gen einlaw_f = empstat if (col10==4 & sex ==2) 

sort hldid einlaw_m einlaw_f
by hldid: replace einlaw_m=einlaw_m[1] if einlaw_m==.& ninlaw==1 & total_inlaw==2

sort hldid einlaw_m einlaw_f
by hldid: replace einlaw_f=einlaw_f[1] if einlaw_f==. & ninlaw==1 & total_inlaw==2

sort hldid einlaw_m einlaw_f
by hldid: replace empsp = einlaw_m if ninlaw==1 & total_inlaw==2 & sex==2 & empsp==-5

sort hldid einlaw_m einlaw_f
by hldid: replace empsp = einlaw_f if ninlaw==1 & total_inlaw==2 & sex==1 & empsp==-5

replace empsp = -7 if civstat == 2

replace empsp=-8 if partid==-8
replace empsp = -7 if partid == -7

recode empsp -7 = -8 if civstat == 1

tab empsp civstat
tab relrefp empsp

recode occupo (0=0)(1=1)(2=2)(3=4)(4/5=5)(6=6)(7=3)(8=8)(9=9)(-7=-7), into(isco1)
tab occupo isco1

gen sector =-5 
replace sector=1 if col26==9
replace sector=2 if col26!=9

gen edcat=-5
replace edcat=1 if (col15>=0 & col15<=2) |(col15==3 & (col16==2|col16==4|col16==5))
replace edcat=2 if (col15==3 & (col16==1|col16==3)) |(col15==4) 
replace edcat=3 if (col15>=5 & col15<=7)

gen rushed=-5
replace rushed=0 if (col19==3|col19==4)
replace rushed=1 if (col19==2)
replace rushed=2 if (col19==1)

//keep only those with diaries
keep if h_suv==1
keep if h_age>9


//Time variables 

//main activities
forvalues i=39(3)468 {
	rename col`i' act`i'
}

local j=1
forvalues i=39(3)468 {
	rename act`i' act`j'
local j=`j'+1
}

//location 
forvalues i=40(3)469 {
	rename col`i' loc`i'
}

local j=1
forvalues i=40(3)469 {
	rename loc`i' loc`j'
local j=`j'+1
}

//with whom
forvalues i=41(3)470 {
	rename col`i' who`i'
}

local j=1
forvalues i=41(3)470 {
	rename who`i' who`j'
local j=`j'+1
}


//Secondary activities
forvalues i=471/614 {
	rename col`i' sec`i'
}

local j=1
forvalues i=471/614 {
	rename sec`i' sec`j'
local j=`j'+1
}


order act* sec* loc* who*

//Make some changes in activities
//1) missing main but second act exists
forvalue i=1/144 {
list hldid id if (act`i'=="S999" & sec`i'!="")
}


gen marker=0
forvalue i=1/144 {
replace marker=1 if (act`i'=="S999" & sec`i'!="")
}


forvalue i=1/144 {
replace act`i'=sec`i' if (act`i'=="S999" & sec`i'!="")
}


replace sec106="" if marker==1 //change the second to main

drop marker


//2) Short gap between activities
forvalue i=2/144 {
	local k=`i'-1
	local j=`i'+1
	list hldid id if (act`i'=="S999" & (act`k'=="S111"|act`j'=="S111"))
	}

	forvalue i=2/144 {
	local k=`i'-1
	local j=`i'+1
	list hldid persid id if (act`i'=="S999" & (act`k'=="S111"|act`j'=="S111"))
	}

forvalue i=2/144 {
	local k=`i'-1
	local j=`i'+1
	replace act`i'= "imp1" if (act`i'=="S999" & (act`k'=="S111"|act`j'=="S111")) ///
	& ((hldid==7686 & id==1)|(hldid==6434 & id==1))
	}

replace act127="imp1" if act128=="imp1"
replace act141="imp1" if act142=="imp1"

forvalue i=1/50 {
list hldid persid id if act`i'=="S999"
}

forvalue i=100/144 {
list hldid persid id if act`i'=="S999"
}

forvalue i=100/144 {
replace act`i'="imps" if act`i'=="S999" & hldid==3199 & persid==2 & id==2
}


//3) Short gap <25 before/after the travel and at home

forvalue i=2/144 {
	local k=`i'-1
	local j=`i'+1
	list hldid persid id if (act`i'=="S999" & (loc`i'==1) &hldid!=4860)
	}

keep countrya survey swave msamp hldid persid id day month year diary sex age act* ///
	sec* loc* who*
	
sort hldid persid id
seq a
rename a unique

forvalues i=1/144 {
	gen timeslot`i'=`i'
	}


reshape long timeslot act sec loc who, i(unique) j(tsl)

//Create time variables
drop secjob sector 

gen epnum=-3
gen clockst=-3
gen start=-3
gen end=-3
gen main=-3
gen secx=-3
gen av==-3
gen inout=-3
gen eloc=-3
gen ict=-3
gen mtrav=-3
gen alone=-3
gen child=-3
gen sppart=-3
gen oad=-3

//create episode level to code meal at work 
sort unique tsl
tsset unique tsl 

gen begin=1 if  act!=  act[_n-1] 
replace begin=1 if unique!= unique[_n-1]

gen finish=1 if  act!=  act[_n+1]
replace finish=1 if unique!= unique[_n+1]

drop if  begin==. &  finish==.

replace begin= tsl if begin==1

replace finish=tsl if finish==1

replace  finish=finish[_n+1] if finish==.

drop if begin==.

//Main activities - imputation

replace main=1 if main==-3 & act=="imp1" //personal care
replace main=2 if main==-3 & (act=="S111"|act=="S112"|act=="S143") //sleep
replace main=3 if main==-3 & act=="imps" 
replace main=4 if main==-3 & act=="S131"|act=="S132"| act=="S141"| act=="S199"
replace main=5 if main==-3 & loc!=1 & (act[_n-1]==("S211")| act[_n-1]=="S212"| ///
		act[_n-1]=="S213"| act[_n-1]=="S214"|act[_n-1]=="S311"|act[_n-1]=="S312" ///
	|act[_n-1]=="S313"|act[_n-1]=="S314"|act[_n-1]=="S319") & ///
	(act=="S122"|act=="S1221"|act=="S1222"|act=="S123") & ///
	(act[_n+1]=="S211"|act[_n+1]=="S212"|act[_n+1]=="S213"|act[_n+1]=="S214"| ///
	act[_n+1]=="S311"|act[_n+1]=="S312" ///
	|act[_n+1]=="S313"|act[_n+1]=="S314"|act[_n+1]=="S319") //eating not at home

replace main=5 if main==-3 & loc!=1 & (act=="S122"|act=="S1221"|act=="S1222"|act=="S123") ///
	& (sec=="S211"|sec=="S212"|sec=="S213"|sec=="S214"|sec=="S216" ///
	| sec=="S311"|sec=="S312"|sec=="S313"|sec=="S314"|sec=="S319") //eating at school or work

replace main =5 if main==-3 & loc!=1 & ((sec=="S122"| sec=="S1221"|sec=="S1222"| ///
				sec=="S123") & (act=="S211"|act=="S212"|act=="S213" ///
						 |act=="S214"|act=="S216" ///
						 | act=="S311"|act=="S312"|act=="S313"| ///
						 act=="S314"|act=="S319"))		

replace main=6 if  main==-3 & ((act=="S121" | act=="S1221" |act=="S1222"| act=="S123")&(loc==1))
replace main=7  if  main==-3 & ((act=="S211"|act=="S214"|act=="S2601"|act=="S2602") & (loc!=1))
replace main=8  if  main==-3 & (act=="S215" | ((act=="S211"|act=="S214"|act=="S2601"|act=="S2602"|act=="S212"|act=="S220"|act=="S230") & (loc==1)))
replace main=9  if main==-3 &  ((act=="S212"|act=="S220" |act=="S230") & (loc!=1))
//no main10 
replace main=11  if  main==-3 & (act=="S822") 
replace main=12  if  main==-3 & (act=="S213")
replace main=13  if  main==-3 & (act=="S216"| act=="S299")
replace main=14  if  main==-3 & (act=="S250")
replace main=15  if  main==-3 & (act=="S311" |act=="S312" |act=="S314"|act=="S319"|act=="S321")
replace main=16  if main==-3 & (act=="S322" |act=="S313")
replace main=17  if main==-3 & (act=="S329" |act=="S721"|act=="S722" |act=="S723"| act=="S724" |act=="S729"|act=="S3301"|act=="S3302")  
replace main=18  if main==-3 & (act=="S411"|act=="S413")
replace main=19  if main==-3 & (act=="S412")
replace main=20  if main==-3 & (act=="S431"|act=="S432"|act=="S433")
replace main=21  if main==-3 & (act=="S421"|act=="S422"|act=="S423")
replace main=22  if main==-3 & (act=="S4411"|act=="S4421")
replace main=23  if main==-3 & (act=="S461"|act=="S462"|act=="S499"|act=="S4431") 
replace main=24  if main==-3 & (act=="S451"|act=="S452"|act=="S4531"|act=="S4532"|act=="S7801"|act=="S7802")
replace main=25  if main==-3 & (act=="S133"|act=="S142")
replace main=26  if main==-3 & (act=="S424"|act=="S4412"|act=="S4422"|act=="S4432"|act=="S463")
//no main27 
replace main=28  if main==-3 & (act=="S511"|act=="S5191"|act=="S5291"|act=="S521") 
replace main=29  if main==-3 & (act=="S522")
replace main=30  if main==-3 & (act=="S5121"|act=="S5122") 
replace main=31  if main==-3 & (act=="S523"|act=="S5192"|act=="S5292")
replace main=32  if main==-3 & (act=="S5301"|act=="S5309"|act=="S5401"|act=="S5409"|act=="S5501"|act=="S5509")
replace main=33  if main==-3 & (act=="S6101"|act=="S6102"|act=="S6109"   ///
| act=="S6211"|act=="S6212"| (act>="S631" & act<="S634") |act=="S639" |act=="S919") 
replace main=34  if main==-3 & (act=="S741"|act=="S742"|act=="S743")
//no main35 
replace main=36  if main==-3 & (act=="S754")
replace main=37  if main==-3 & (act=="S751"|act=="S752")
replace main=38  if main==-3 & (act=="S753"|act=="S755")
replace main=39  if main==-3 & (act=="S775"| ((act=="S121" |act=="S1221"|act=="S1222"|act=="S123")&(loc!=1)))
replace main=40  if main==-3 & (((act=="S715"| act=="S719") & (loc!=1)))
replace main=41  if main==-3 & (act=="S999" & loc!=1)
replace main=42  if main==-3 &  (act=="S763"|act=="S764"|act=="S766")
replace main=43  if main==-3 & (act=="S761"|act=="S762")
//no main44 
replace main=45  if main==-3 & (act=="S765")
replace main=46  if main==-3 & (act=="S240")
//no main47
replace main=48  if main==-3 & (((act=="S712"|act=="S713")|((act=="S719"|act=="S715")&(loc==1))))
replace main=49  if main==-3 & (act=="S711")
replace main=50  if main==-3 & (act=="S773")
replace main=51  if main==-3 & (act=="S799")
//no main52 and main53
replace main=54  if main==-3 & (act=="S774"|act=="S425")
replace main=55  if main==-3 & (act=="S776"|act=="S777")
replace main=56  if main==-3 & (act=="S771"|act=="S731"|act=="S732")
replace main=57  if main==-3 & (act=="S736")
replace main=58  if main==-3 & (act=="S735")
replace main=59  if main==-3 & (act=="S733"|act=="S734")
replace main=60  if main==-3 & (act=="S772")
replace main=61  if main==-3 & (act=="S714"|act=="S7371"|act=="S7372")
//no main62
replace main=63  if main==-3 & (act=="S821")
replace main=64  if main==-3 &  (act=="S831"|act=="S8721")
replace main=65  if main==-3 & (act=="S861")
replace main=66  if main==-3 & (act=="S851")
replace main=67  if main==-3 & (act=="S811"|act=="S841")
replace main=68  if main==-3 & (act=="S800"|act=="S871"|act=="S8722"|act=="S891")
replace main=69  if main==-3 & ( act=="S999" & loc==1) 

//secondary actvities
replace secx=2 if secx==-3 & (sec=="S111"|sec=="S112"|sec=="S143")
replace secx=4 if secx==-3 & (sec=="S131"|sec=="S132"| sec=="S141"| sec=="S199")
replace secx=6 if secx==-3 & (((sec=="S121" | sec=="S1221" |sec=="S1222"| sec=="S123")&(loc==1)))
replace secx=7  if secx==-3 & ((sec=="S211"|sec=="S214"|sec=="S2601"|sec=="S2602") & (loc!=1))
replace secx=8  if secx==-3 & (sec=="S215" | ((sec=="S211"|sec=="S214"|sec=="S2601"|sec=="S2602"|sec=="S212"|sec=="S220"|sec=="S230") & (loc==1))) 
replace secx=9  if secx==-3 & (((sec=="S212"|sec=="S220" |sec=="S230") & (loc!=1)))
replace secx=11  if secx==-3 & (sec=="S822")
replace secx=12  if secx==-3 & (sec=="S213")
replace secx=13  if secx==-3 & (sec=="S216"| sec=="S299")
replace secx=14  if secx==-3 & (sec=="S250")
replace secx=15  if secx==-3 & (sec=="S311" |sec=="S312" |sec=="S314"|sec=="S319"|sec=="S321")
replace secx=16  if secx==-3 & (sec=="S322" |sec=="S313") 
replace secx=17  if secx==-3 & (sec=="S329" |sec=="S721"|sec=="S722" |sec=="S723"| sec=="S724" |sec=="S729"|sec=="S3301"|sec=="S3302")   
replace secx=18  if secx==-3 & (sec=="S411"|sec=="S413")
replace secx=19  if secx==-3 & (sec=="S412")
replace secx=20  if secx==-3 & (sec=="S431"|sec=="S432"|sec=="S433")
replace secx=21  if secx==-3 & (sec=="S421"|sec=="S422"|sec=="S423")
replace secx=22  if secx==-3 & (sec=="S4411"|sec=="S4421")
replace secx=23  if secx==-3 & (sec=="S461"|sec=="S462"|sec=="S499"|sec=="S4431") 
replace secx=24  if secx==-3 & (sec=="S451"|sec=="S452"|sec=="S4531"|sec=="S4532"|sec=="S7801"|sec=="S7802")
replace secx=25  if secx==-3 & (sec=="S133"|sec=="S142")
replace secx=26  if secx==-3 & (sec=="S424"|sec=="S4412"|sec=="S4422"|sec=="S4432"|sec=="S463")
replace secx=28  if secx==-3 & (sec=="S511"|sec=="S5191"|sec=="S5291"|sec=="S521")
replace secx=29  if secx==-3 & (sec=="S522")
replace secx=30  if secx==-3 & (sec=="S5121"|sec=="S5122") 
replace secx=31  if secx==-3 & (sec=="S523"|sec=="S5192"|sec=="S5292")
replace secx=32  if secx==-3 & (sec=="S5301"|sec=="S5309"|sec=="S5401"|sec=="S5409"|sec=="S5501"|sec=="S5509")
replace secx=33  if secx==-3 & (sec=="S6101"|sec=="S6102"|sec=="S6109"   ///
| sec=="S6211"|sec=="S6212"| (sec>="S631" & sec<="S634") |sec=="S639" |sec=="S919")
replace secx=34  if secx==-3 & (sec=="S741"|sec=="S742"|sec=="S743")
replace secx=36  if secx==-3 & (sec=="S754")
replace secx=37  if secx==-3 & (sec=="S751"|sec=="S752")
replace secx=38  if secx==-3 & (sec=="S753"|sec=="S755")
replace secx=39  if secx==-3 & (sec=="S775"| ((sec=="S121" |sec=="S1221"|sec=="S1222"|sec=="S123")&(loc!=1)))
replace secx=40  if secx==-3 & ((sec=="S715"| sec=="S719") & (loc!=1))
replace secx=41  if secx==-3 & (sec=="S999" & loc!=1)
replace secx=42  if secx==-3 & (sec=="S763"|sec=="S764"|sec=="S766")
replace secx=43  if secx==-3 & (sec=="S761"|sec=="S762")
replace secx=45  if secx==-3 & (sec=="S765")
replace secx=46  if secx==-3 & (sec=="S240")
replace secx=48  if secx==-3 & (((sec=="S712"|sec=="S713")|((sec=="S719"|sec=="S715")&(loc==1))))
replace secx=49  if secx==-3 & sec=="S711"
replace secx=50  if secx==-3 & sec=="S773"
replace secx=51  if secx==-3 & sec=="S799"
replace secx=54  if secx==-3 & (sec=="S774"|sec=="S425")
replace secx=55  if secx==-3 & (sec=="S776"|sec=="S777")
replace secx=56  if secx==-3 & (sec=="S771"|sec=="S731"|sec=="S732")
replace secx=57  if secx==-3 & (sec=="S736")
replace secx=58  if secx==-3 & sec=="S735"
replace secx=59  if secx==-3 & (sec=="S733"|sec=="S734")
replace secx=60  if secx==-3 & sec=="S772"
replace secx=61  if secx==-3 & (sec=="S714"|sec=="S7371"|sec=="S7372")
replace secx=63  if secx==-3 & sec=="S821"
replace secx=64  if secx==-3 & (sec=="S831"|sec=="S8721")
replace secx=65  if secx==-3 & sec=="S861"
replace secx=66  if secx==-3 & sec=="S851"
replace secx=67  if secx==-3 & (sec=="S811"|sec=="S841")
replace secx=68  if secx==-3 & (sec=="S800"|sec=="S871"|sec=="S8722"|sec=="S891")
replace secx=69  if secx==-3 & (sec=="S999" & loc==1)

//AV -- 41 MTUS main categories
replace av=1 if av==-3 & ((act=="S211"| act=="S213"| act=="S214"| act=="S2601"|  act=="S2602"| act=="S216"| act=="S299") & (loc!=1))
replace av=2 if av==-3 & (((loc==1) & (act=="S211"| act=="S213"| act=="S214"| act=="S2601"|  act=="S2602"| act=="S216"| act=="S299"))| act=="S215"|act=="S250")
replace av=3 if av==-3 & (act=="S212"|act=="S220"|act=="S230")
replace av=4 if av==-3 & (act=="S311"|act=="S312"|act=="S313"|act=="S314"|act=="S319"|act=="S321"|act=="S723"|act=="S724")
replace av=5 if av==-3 & (act=="S821"|act=="S822"|act=="S831")
replace av=6 if av==-3 & (act=="S411"|act=="S412"|act=="S413")
replace av=7 if av==-3 & (act=="S421"|act=="S422"|act=="S423"|act=="S431"|act=="S432"|act=="S433"|act=="S499")
replace av=8 if av==-3 & (act=="S4411"|act=="S4421"|act=="S4431"|act=="S461"|act=="S462"|act=="S5301"|act=="S5309"|act=="S5401"|act=="S5409"|act=="S5501"|act=="S5509"|act=="S6101"|act=="S6102"|act=="S6109")
replace av=9 if av==-3 & act=="S240"
replace av=10 if av==-3 & (act=="S3301"|act=="S3302"|act=="S424"|act=="S4412"|act=="S4422"|act=="S4432"|act=="S451"|act=="S452"|act=="S4531"|act=="S4532"|act=="S7801"|act=="S7802"|act=="S463")
replace av=11 if av==-3 & (act=="S511"|act=="S5121"|act=="S5122"|act=="S5191"|act=="S5291"|act=="S5192"|act=="S5292"|act=="S521"|act=="S522"|act=="S523")
replace av=12 if av==-3 & (act=="S811"|act=="S841"|act=="S851")
replace av=13 if av==-3 & (act=="S131"|act=="S132"|act=="S199")
replace av=14 if av==-3 & (act=="S133"|act=="S142"|act=="S141")
replace av=15 if av==-3 & (((act=="S121"|act=="S1221"|act=="S1222"|act=="S123") & (loc==1)))
replace av=16 if av==-3 &(act=="S111"|act=="S112"||act=="S143")
replace av=17 if av==-3 & (act=="S800"|act=="S861"|act=="S871"|act=="S8721"|act=="S8722"|act=="S891")
replace av=18 if av==-3 & (act=="S753"|act=="S755"|act=="S765")
replace av=19 if av==-3 & (act=="S763"|act=="S764"|act=="S766")
replace av=20 if av==-3 & act=="S754"
replace av=21 if av==-3 & (act=="S761"|act=="S762")
replace av=22 if av==-3 & (act=="S741"|act=="S742"|act=="S743")
replace av=23 if av==-3 & (act=="S6211"|act=="S6212"|act=="S631"|act=="S632"|act=="S633"|act=="S634"|act=="S639"|act=="S919")
replace av=24 if av==-3 & (act=="S751"|act=="S752")
replace av=25 if av==-3 & (((act=="S715"|act=="S719") & (loc!=1)))
//no av26
replace av=27 if av==-3 & act=="S775"
replace av=28 if av==-3 & (((act=="S121"|act=="S1221"|act=="S1222"|act=="S123")&(loc!=1)))
replace av=29 if av==-3 & ((act=="S712"|act=="S713") & loc!=1)
replace av=30 if av==-3 & act=="S735"
replace av=31 if av==-3 & (act=="S733"|act=="S734")
replace av=32 if av==-3 & act=="S736"
replace av=33 if av==-3 & (act=="S322"|act=="S329"|act=="S721"|act=="S722"|act=="S729")
replace av=34 if av==-3 & act=="S771"
replace av=35 if av==-3 & (act=="S731"|act=="S732")
replace av=36 if av==-3 & (act=="S776"|act=="S777")
replace av=37 if av==-3 & act=="S711"
replace av=38 if av==-3 & (((act=="S712"|act=="S713"|act=="S719"|act=="S715")&(loc==1)))
replace av=39 if av==-3 & act=="S425"
replace av=40 if av==-3 & (act=="S714"|act=="S7371"|act=="S7372"|act=="S772"|act=="S773"|act=="S774"|act=="S799") 
replace av=41 if av==-3 & act=="S999"

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

//Location 
replace eloc=1 if loc==1 & eloc==-3 
//no eloc=2 
replace eloc=3 if eloc==-3 & loc==0 & (main==7 | main==9 |main==12| main==13) 
replace eloc=4 if eloc==-3 & loc==0 & (act=="S311"|act=="S312"|act=="S313"|act=="S314"|act=="S319"|act=="S321")
replace eloc=5 if eloc==-3 & loc==0 & (act=="S133"|act=="S142"|act=="S2601"|act=="S3301"| ///
							act=="S4422"|act=="S451"|act=="S452"|act=="S463"| ///
							act=="S7801")
replace eloc=6 if eloc==-3 &  loc==0 & main==39
replace eloc=7 if eloc==-3 & loc==0 & (act=="S741"|act=="S742"|act=="S743")
replace eloc=8 if eloc==-3 & loc>=2 & loc<=9
replace eloc=9 if eloc==-3 

//inout
replace inout=3 if inout==-3& loc>=3 & loc<=9
replace inout=2 if inout==-3& (loc==2 | act=="S715"| act=="S76" | act=="S761" ///
						  | act=="S762"| act=="S764"| act=="S765" ///
						  | act=="S766")
replace inout=1 if inout==-3 & (loc==1 | act=="S111"| act=="S112"| act=="S131" ///
							| act=="S132" |act=="S133"| act=="S141"|act=="S142" ///
							|act=="S143"|act=="S199"|act=="S215"|eloc==4 ///
							|main==18| main==19| main==20|main==21|main==23|main==39 ///
							|act=="S511"| act=="S5121"| act=="S521"| act=="S751" ///
							|act=="S752"| act=="S753"| act=="S741"|act=="S742"|act=="S743")
replace inout=-8 if inout==-3

//mode of transportation
replace mtrav=-8 if mtrav==-3& act=="S999"
replace mtrav=-7 if mtrav==-3 & (loc==1|inout==1)
replace mtrav=1 if mtrav==-3 & (loc==5|loc==6|loc==7)
replace mtrav=2 if mtrav==-3 & (loc==3|loc==4|loc==8)
replace mtrav=3 if mtrav ==-3 & ((loc==2)|(act=="S761")|(act=="S762"))
replace mtrav=5 if mtrav==-3 & (loc==9)
replace mtrav=-7 if mtrav==-3 & eloc!=8

//With whom
//spouse present
replace sppart=1 if who==2
replace sppart=0 if who!=2
replace sppart=1 if act=="S530"| act=="S5301"| act=="S5309"

//child present
replace child=1 if child==-3& who==3
replace child=1 if child==-3 & (act=="S511"|act=="S512"|act=="S5121"|act=="S5122"|act=="S519" ///
								|act=="S5191"|act=="S5192"|act=="S521"|act=="S522" ///
								|act=="S529"| act=="S5291"|act=="S5292")
replace child=0 if child==-3

//other adult present
replace oad=1 if who==4|who==5|sppart==1
replace oad=1 if act=="S540" | act=="S5401"|act=="S5409"
replace oad=0 if oad==-3 
					
//alone
replace alone=1 if alone==-3 & (who==1)
replace alone=0 if alone==-3 & who!=1
 
gen alone2=-3
replace alone2=1 if (who==1 & sppart==0 & child==0 & oad==0)
replace alone2=0 if alone2==-3 							

rename alone alone_owndef
rename alone2 alone


//use of ict during activity
gen ict=-3
replace ict=1 if act=="S714"|act=="S737"|act=="S7371"|act=="7372"|act=="S772"|act=="722"
replace ict=0 if ict==-3
				
//episode number
bysort unique: replace epnum=_n

//clockst, start, end time

replace start=(begin-1)*10
order unique tsl start end begin finish
replace end=finish*10

gen cl_temp=0
replace cl_temp=start 
gen cl_hour=trunc(cl_temp/60)
gen cl_min=((cl_temp-(cl_hour*60))*0.01)
replace clockst=cl_hour+cl_min
format clockst %12.2f

gen time = end-start

//bad case
by unique: egen eatdrkt = total(time) if inlist(av, 6, 15, 27, 28, 29, 38)
by unique: egen rstt = total(time) if inlist(av, 16, 36)
by unique: egen pcaret = total(time) if inlist(av, 13, 14)

by unique: egen eatdrk = max(eatdrkt)
by unique: egen rst = max(rstt)
by unique: egen pcare = max(pcaret)

replace eatdrk = 0 if eatdrk ==.
replace rst = 0 if rst ==.
replace pcare = 0 if pcare ==.

//noweight
gen nowght = -5
replace nowght = 1 if main==69 & time>90
replace nowght = 1 if epnumx<7 
replace nowght = 1 if (eatdrk==0 & rst==0 & pcare==0)

tab nowght
replace nowght=0 if nowght==-5

//original weight
rename col38 ocombwt //individual weight

//variables added
gen region = -9
gen ethnic = -9
gen migrantd = -9
gen migrantm = -9
gen migrantf = -9

//checks
tab mtrav
tab inout if mtrav==-8
list main if inout==1 & mtrav==-8

list loc if inout==1 & mtrav==-8
replace mtrav=-7 if inout==1 & mtrav==-8

tab av
list main if av==-3
replace av=16 if av==-3 & main==3
replace av=13 if av==-3 & main==1 

replace secx=69 if secx==-3

list main if hldid==3006 & persid==3 & id==1