******************************************************************************** /* Citation: Oxford Poverty and Human Development Initiative (OPHI), University of Oxford. 2018 Global Multidimensional Poverty Index - South Sudan MICS 2010 [STATA do-file]. Available from OPHI website: http://ophi.org.uk/ For further queries, contact: ophi@qeh.ox.ac.uk */ ******************************************************************************** clear all set more off set maxvar 10000 set mem 500m cap log close *** Working Folder Path *** global path_in "T:/GMPI 2.0/data/South Sudan MICS 2010" global path_out "D:/pov" global path_logs "D:/logs" global path_ado "D:/ado" *** Log file *** log using "$path_logs/ssd_mics10_dataprep.log", replace ******************************************************************************** *** SOUTH SUDAN MICS 2010 *** ******************************************************************************** ******************************************************************************** *** Step 1: Data preparation *** Selecting main variables from CH, WM, HH & MN recode & merging with HL recode ******************************************************************************** ******************************************************************************** *** Step 1.1 CH - CHILDREN's RECODE (under 5) ******************************************************************************** use "$path_in/ch.dta", clear rename _all, lower *** Generate individual unique key variable required for data merging *** hh1=cluster number; *** hh2=household number; *** ln=child's line number in household gen double ind_id = hh1*100000 + hh2*100 + ln format ind_id %20.0g label var ind_id "Individual ID" duplicates report ind_id //No duplicates gen child_CH=1 //Generate identification variable for observations in CH recode /* For this part of the do-file we use the WHO Anthro and macros. This is to calculate the z-scores of children under 5. Source of ado file: http://www.who.int/childgrowth/software/en/ */ *** Next, indicate to STATA where the igrowup_restricted.ado file is stored: adopath + "$path_ado/igrowup_stata" *** We will now proceed to create three nutritional variables: *** weight-for-age (underweight), *** weight-for-height (wasting) *** height-for-age (stunting) /* We use 'reflib' to specify the package directory where the .dta files containing the WHO Child Growth Standards are stored. Note that we use strX to specify the length of the path in string. If the path is long, you may specify str55 or more, so it will run. */ gen str100 reflib = "$path_ado/igrowup_stata" lab var reflib "Directory of reference tables" /* We use datalib to specify the working directory where the input STATA dataset containing the anthropometric measurement is stored. */ gen str100 datalib = "$path_out" lab var datalib "Directory for datafiles" /* We use datalab to specify the name that will prefix the output files that will be produced from using this ado file (datalab_z_r_rc and datalab_prev_rc)*/ gen str30 datalab = "children_nutri_ssd" lab var datalab "Working file" *** Next check the variables that WHO ado needs to calculate the z-scores: *** sex, age, weight, height, measurement, oedema & child sampling weight *** Variable: SEX *** tab hl4, miss //"1" for male ;"2" for female tab hl4, nol clonevar gender = hl4 desc gender tab gender *** Variable: AGE *** clonevar age_months = cage desc age_months summ age_months gen str6 ageunit = "months" lab var ageunit "Months" *** Variable: BODY WEIGHT (KILOGRAMS) *** /*Note: In the South Sudan MICS 2010 dataset, some 3,305 (33%) of the children under 5 had missing weight data.*/ codebook an3, tab (999) clonevar weight = an3 replace weight = . if an3>=99 tab an2 an3 if an3>=99 | an3==., miss //an2: result of the measurement tab uf9 if an2==. & an3==. desc weight summ weight *** Variable: HEIGHT (CENTIMETERS) /*Note: In the South Sudan MICS 2010 dataset, some 3,715 (37%) of the children under 5 had missing height data.*/ codebook an4, tab (999) clonevar height = an4 replace height = . if an4>=999 tab an2 an4 if an4>=999 | an4==., miss desc height summ height *** Variable: MEASURED STANDING/LYING DOWN codebook an4a gen measure = "l" if an4a==1 //Child measured lying down replace measure = "h" if an4a==2 //Child measured standing up replace measure = " " if an4a==9 | an4a==0 | an4a==. //Replace with " " if unknown desc measure tab measure *** Variable: OEDEMA *** lookfor oedema gen oedema = "n" if an5==2 replace oedema = "y" if an5==1 replace oedema = " " if an5==3 | an5==4 | an5==9 | an5==. desc oedema *** Variable: INDIVIDUAL CHILD SAMPLING WEIGHT *** gen sw = chweight desc sw summ sw /*We now run the command to calculate the z-scores with the adofile */ igrowup_restricted reflib datalib datalab gender age_months ageunit weight /// height measure oedema sw /*We now turn to using the dta file that was created and that contains the calculated z-scores */ use "$path_out/children_nutri_ssd_z_rc.dta", clear gen underweight = (_zwei < -2.0) replace underweight = . if _zwei == . | _fwei==1 lab var underweight "Child is undernourished (weight-for-age) 2sd - WHO" tab underweight, miss gen stunting = (_zlen < -2.0) replace stunting = . if _zlen == . | _flen==1 lab var stunting "Child is stunted (length/height-for-age) 2sd - WHO" tab stunting, miss gen wasting = (_zwfl < - 2.0) replace wasting = . if _zwfl == . | _fwfl == 1 lab var wasting "Child is wasted (weight-for-length/height) 2sd - WHO" tab wasting, miss count if _fwei==1 | _flen==1 /* Note: in the context of South Sudan MICS 2010, 305 children were replaced as '.' because they have extreme z-scores that are biologically implausible.*/ //Retain relevant variables: keep ind_id child_CH ln underweight stunting wasting order ind_id child_CH ln underweight stunting wasting sort ind_id duplicates report ind_id //Erase files from folder: erase "$path_out/children_nutri_ssd_z_rc.xls" erase "$path_out/children_nutri_ssd_prev_rc.xls" erase "$path_out/children_nutri_ssd_z_rc.dta" //Save a temp file for merging with HL: save "$path_out/SSD10_CH.dta", replace /*Note: Some 37% of children under 5 have either missing info on height or weight. The country report confirmed the high missing anthropometric information(p.27). The report further stated that the unusual high missing values may be because of the failure of enumerators to patiently collect the required information from children under 5 (p.9)*/ ******************************************************************************** *** Step 1.2 BH - BIRTH RECODE *** (All females 15-49 years who ever gave birth) ******************************************************************************** /*The purpose of step 1.2 is to identify children of any age who died in the last 5 years prior to the survey date.*/ use "$path_in/bh.dta", clear rename _all, lower *** Generate individual unique key variable required for data merging using: *** hh1=cluster number; *** hh2=household number; *** wm4=women's line number. gen double ind_id = hh1*100000 + hh2*100 + ln format ind_id %20.0g label var ind_id "Individual ID" desc bh4c bh9c gen date_death = bh4c + bh9c //Date of death = date of birth (bh4c) + age at death (bh9c) gen mdead_survey = wdoi-date_death //Months dead from survey = Date of interview (wdoi) - date of death replace mdead_survey = . if (bh9c==0 | bh9c==.) & bh5==1 /*Replace children who are alive as '.' to distinguish them from children who died at 0 months */ gen ydead_survey = mdead_survey/12 //Years dead from survey codebook bh5, tab (10) //bh5 - Child still alive: 1=Yes; 2=No gen child_died = 1 if bh5==2 //1=child dead; 0=child alive replace child_died = 0 if bh5==1 replace child_died = . if bh5==. label define lab_died 0"child is alive" 1"child has died" label values child_died lab_died tab bh5 child_died, miss bysort ind_id: egen tot_child_died = sum(child_died) //For each woman, sum the number of children who died bysort ind_id: egen tot_child_died_5y=sum(child_died) if ydead_survey<=5 /*For each woman, sum the number of children who died in the past 5 years prior to the interview date */ replace tot_child_died_5y=0 if tot_child_died_5y==. & tot_child_died>=0 & tot_child_died<. /*All children who are alive and died longer than 5 years from the interview date are replaced as '0'*/ replace tot_child_died_5y=. if child_died==1 & ydead_survey==. //Replace as '.' if there is no information on when the child died tab tot_child_died tot_child_died_5y, miss bysort ind_id: egen child_died_per_wom = max(tot_child_died) lab var child_died_per_wom "Total child death for each women (birth recode)" bysort ind_id: egen child_died_per_wom_5y = max(tot_child_died_5y) lab var child_died_per_wom_5y "Total child death for each women in the last 5 years (birth recode)" //Keep one observation per women bysort ind_id: gen id=1 if _n==1 keep if id==1 drop id duplicates report ind_id gen women_BH = 1 //Identification variable for observations in BH recode //Retain relevant variables keep ind_id hh1 hh2 women_BH child_died_per_wom child_died_per_wom_5y order ind_id hh1 hh2 women_BH child_died_per_wom child_died_per_wom_5y sort ind_id //Save a temp file for merging with HL: save "$path_out/SSD10_BH.dta", replace ******************************************************************************** *** Step 1.3 WM - WOMEN's RECODE *** (All eligible females 15-49 years in the household) ******************************************************************************** use "$path_in/wm.dta", clear rename _all, lower *** Generate individual unique key variable required for data merging *** hh1=cluster number; *** hh2=household number; *** ln=respondent's line number gen double ind_id = hh1*100000 + hh2*100 + ln format ind_id %20.0g label var ind_id "Individual ID" duplicates report ind_id gen women_WM =1 //Identification variable for observations in WM recode tab wb2, miss //Women in the sample: 15-49 years lookfor marital codebook mstatus ma6, tab (10) tab mstatus ma6, miss gen marital = 1 if mstatus == 3 & ma6==. //1: Never married replace marital = 2 if mstatus == 1 & ma6==. //2: Currently married replace marital = 3 if mstatus == 2 & ma6==1 //3: Widowed replace marital = 4 if mstatus == 2 & ma6==2 //4: Divorced replace marital = 5 if mstatus == 2 & ma6==3 //5: Separated/not living together label define lab_mar 1"never married" 2"currently married" 3"widowed" /// 4"divorced" 5"not living together" label values marital lab_mar label var marital "Marital status of household member" tab marital, miss tab ma6 marital, miss tab mstatus marital, miss //Retain relevant variables: keep wm7 cm1 cm8 cm9a cm9b ind_id women_WM marital order wm7 cm1 cm8 cm9a cm9b ind_id women_WM marital sort ind_id //Save a temp file for merging with HL: save "$path_out/SSD10_WM.dta", replace ******************************************************************************** *** Step 1.4 MN - MEN'S RECODE ***(All eligible man: 15-54 / 15-59 years in the household) ******************************************************************************** /*The male data file was not made available for South Sudan MICS 2010. The country report indicate that the response for the men’s module was exceedingly low, as their overall response rate is 48. As such, all analysis on men was excluded (p.6).*/ ******************************************************************************** *** Step 1.5 HH - HOUSEHOLD RECODE ***(All households interviewed) ******************************************************************************** use "$path_in/hh.dta", clear rename _all, lower *** Generate individual unique key variable required for data merging *** hh1=cluster number; *** hh2=household number; gen double hh_id = hh1*100 + hh2 format hh_id %20.0g lab var hh_id "Household ID" duplicates report hh_id //Save a temp file for merging with HL: save "$path_out/SSD10_HH.dta", replace ******************************************************************************** *** Step 1.6 HL - HOUSEHOLD MEMBER ******************************************************************************** use "$path_in/hl.dta", clear rename _all, lower gen cty = "South Sudan" gen ccty = "SSD" gen year = "2010" gen survey = "MICS" gen ccnum = 728 *** Generate a household unique key variable at the household level using: ***hh1=cluster number ***hh2=household number gen double hh_id = hh1*100 + hh2 format hh_id %20.0g label var hh_id "Household ID" *** Generate individual unique key variable required for data merging using: *** hh1=cluster number; *** hh2=household number; *** hl1=respondent's line number. gen double ind_id = hh1*100000 + hh2*100 + hl1 format ind_id %20.0g label var ind_id "Individual ID" duplicates report ind_id sort ind_id ******************************************************************************** *** Step 1.7 DATA MERGING ******************************************************************************** *** Merging BR Recode ***************************************** merge 1:1 ind_id using "$path_out/SSD10_BH.dta" drop _merge erase "$path_out/SSD10_BH.dta" *** Merging WM Recode ***************************************** merge 1:1 ind_id using "$path_out/SSD10_WM.dta" tab hl7, miss gen temp = (hl7>0) tab women_WM temp, miss col tab wm7 if temp==1 & women_WM==., miss //Total of eligible women not interviewed drop temp drop _merge erase "$path_out/SSD10_WM.dta" *** Merging HH Recode ***************************************** merge m:1 hh_id using "$path_out/SSD10_HH.dta" tab hh9 if _m==2 drop if _merge==2 //Drop households that were not interviewed drop _merge erase "$path_out/SSD10_HH.dta" *** Merging CH Recode ***************************************** merge 1:1 ind_id using "$path_out/SSD10_CH.dta" drop _merge erase "$path_out/SSD10_CH.dta" sort ind_id ******************************************************************************** *** Step 1.8 CONTROL VARIABLES ******************************************************************************** /* Households are identified as having 'no eligible' members if there are no applicable population, that is, children 0-5 years, adult women 15-49 years or adult men. These households will not have information on relevant indicators of health. As such, these households are considered as non-deprived in those relevant indicators.*/ *** No Eligible Women 15-49 years ***************************************** gen fem_eligible = (hl7>0) if hl7!=. bys hh_id: egen hh_n_fem_eligible = sum(fem_eligible) gen no_fem_eligible = (hh_n_fem_eligible==0) //Takes value 1 if the household had no eligible females for an interview lab var no_fem_eligible "Household has no eligible women" tab no_fem_eligible, miss *** No Eligible Men ***************************************** /*Since there is no male data file, this variable is generated as an empty variable */ gen no_male_eligible = . lab var no_male_eligible "Household has no eligible man" tab no_male_eligible, miss *** No Eligible Children 0-5 years ***************************************** gen child_eligible = (hl9>0 | child_CH==1) bys hh_id: egen hh_n_children_eligible = sum(child_eligible) //Number of eligible children for anthropometrics gen no_child_eligible = (hh_n_children_eligible==0) //Takes value 1 if there were no eligible children for anthropometrics lab var no_child_eligible "Household has no children eligible" tab no_child_eligible, miss *** No Eligible Women and Men *********************************************** /*Since there is no male data file, this variable is generated as an empty variable */ gen no_adults_eligible = . lab var no_adults_eligible "Household has no eligible women or men" tab no_adults_eligible, miss *** No Eligible Children and Women *********************************************** /*NOTE: In the DHS datasets, we use this variable as a control variable for the nutrition indicator if nutrition data is present for children and women. However, in MICS, we do NOT use this as a control variable. This is because nutrition data is only collected from children. However, we continue to generate this variable in this do-file so as to be consistent*/ gen no_child_fem_eligible = (no_child_eligible==1 & no_fem_eligible==1) lab var no_child_fem_eligible "Household has no children or women eligible" tab no_child_fem_eligible, miss *** No Eligible Women, Men or Children *********************************************** /*NOTE: In the DHS datasets, we use this variable as a control variable for the nutrition indicator if nutrition data is present for children, women and men. However, in MICS, we do NOT use this as a control variable. This is because nutrition data is only collected from children. However, we continue to generate this variable in this do-file so as to be consistent Since there is no male data file, this variable is generated as an empty variable */ gen no_eligibles = . lab var no_eligibles "Household has no eligible women, men, or children" tab no_eligibles, miss *** No Eligible Subsample ***************************************** /*Note that the MICS surveys do not collect hemoglobin data. As such, this variable takes missing value. However, we continue to generate this variable in this do-file so as to be consistent*/ gen no_hem_eligible = . lab var no_hem_eligible "Household has no eligible individuals for hemoglobin measurements" drop fem_eligible hh_n_fem_eligible child_eligible hh_n_children_eligible sort hh_id ******************************************************************************** *** Step 1.9 RENAMING DEMOGRAPHIC VARIABLES *** ******************************************************************************** //Sample weight clonevar weight = hhweight label var weight "Sample weight" //Area: urban or rural desc hh6 clonevar area = hh6 replace area=0 if area==2 label define lab_area 1 "urban" 0 "rural" label values area lab_area label var area "Area: urban-rural" //Relationship to the head of household desc hl3 clonevar relationship = hl3 codebook relationship, tab (20) recode relationship (1=1)(2=2)(15=2)(3=3)(13=3)(14=5)(4/12=4)(16=4) /*Please ensure recoding follows these 6 broad categories: 1"head" 2"spouse" 3"child" 4"extended family" 5"not related" 6"maid" */ label define lab_rel 1"head" 2"spouse" 3"child" 4"extended family" /// 5"not related" 6"maid" label values relationship lab_rel label var relationship "Relationship to the head of household" tab hl3 relationship, miss //Sex of household member codebook hl4 clonevar sex = hl4 label var sex "Sex of household member" //Age of household member codebook hl6, tab (100) clonevar age = hl6 replace age = . if age>=98 label var age "Age of household member" //Age group recode age (0/4 = 1 "0-4")(5/9 = 2 "5-9")(10/14 = 3 "10-14") /// (15/17 = 4 "15-17")(18/59 = 5 "18-59")(60/max=6 "60+"), gen(agec7) lab var agec7 "age groups (7 groups)" recode age (0/9 = 1 "0-9") (10/17 = 2 "10-17")(18/59 = 3 "18-59") /// (60/max=4 "60+"), gen(agec4) lab var agec4 "age groups (4 groups)" //Total number of hh members in the household gen member = 1 bysort hh_id: egen hhsize = sum(member) label var hhsize "Household size" tab hhsize, miss compare hhsize hh11 drop member //Religion of the household head lookfor religion gen religion_hh = . label var religion_hh "Religion of household head" //Ethnicity of the household head lookfor ethnicity gen ethnic_hh = . label var ethnic_hh "Ethnicity of household head" //Subnational region /*Note: The primary objective of the sample design for the South Sudan MICS 2010 was to produce statistically reliable estimates for a large number of indicators on the situation of children and women at the national level, for urban and rural areas, and for each of the ten states of the country (Upper Nile, Jonglei, Unity, Warap, Northern Bahr El Ghazal, Western Bahr El Ghazal, Lakes, Western Equatoria, Central Equatoria, and Eastern Equatoria)(p.164). It should be noted that the country report cautions that across the 10 States, women’s response rates, except Northern Bahr el Ghazal, are below 85 percent. The results for these States should thus be interpreted with some caution, as their response rates are low. The response rates for the children under five years of age in 5 of the 10 States were equally low. These are Western Equatoria, Central Equatoria, Unity, Upper Nile and Lakes States. These results are low, and therefore interpretation in these States should also be handled with caution. Response rates for urban and rural areas are also below 85 percent; this as well requires some caution in the interpretation of any results (see page 6 of report). */ codebook hh7, tab (99) decode hh7, gen(temp) replace temp = proper(temp) encode temp, gen(region) lab var region "Region for subnational decomposition" tab hh7 region, miss drop temp ******************************************************************************** *** Step 2 Data preparation *** *** Standardization of the 10 Global MPI indicators *** Identification of non-deprived & deprived individuals ******************************************************************************** ******************************************************************************** *** Step 2.1 Years of Schooling *** ******************************************************************************** tab ed4b ed4a, miss tab age ed6a if ed5==1, miss clonevar edulevel = ed4a //Highest educational level attended replace edulevel = . if ed4a==. |ed4a==98 | ed4a==99 //ed4a=98/99 are missing values replace edulevel = 0 if ed3==2 //Those who never attended school are replaced as '0' label var edulevel "Highest educational level attended" clonevar eduhighyear = ed4b //Highest grade of education completed replace eduhighyear = . if ed4b==. | ed4b==97 | ed4b==98 | ed4b==99 //ed4b=97/98/99 are missing values replace eduhighyear = 0 if ed3==2 //Those who never attended school are replaced as '0' lab var eduhighyear "Highest year of education completed" *** Cleaning inconsistencies replace eduhighyear = 0 if age<10 /*The variable "eduhighyear" was replaced with a '0' given that the criteria for this indicator is household member aged 10 years or older */ replace eduhighyear = 0 if edulevel<1 *** Now we create the years of schooling tab eduhighyear edulevel, miss gen eduyears = eduhighyear replace eduyears = 0 if edulevel<=1 & eduhighyear==. /*Assuming 0 year if they only attend preschool or primary but the last year is unknown*/ replace eduyears = eduhighyear + 8 if edulevel == 2 | edulevel==3 replace eduyears = 8 if (edulevel == 2 | edulevel==3) & eduhighyear == . /*29 cases with missing information were given at least the years of education of primary level */ replace eduyears = 11 if edulevel >3 & edulevel< 7 /*eduhighyears is missing for all post-secondary, so 8+4 for post secondary education */ replace eduyears = 0 if edulevel == 7| edulevel==8 //informal education is replaced as 0 years of education *** Checking for further inconsistencies replace eduyears = . if age<=eduyears & age>0 /*There are cases in which the years of schooling are greater than the age of the individual. This is clearly a mistake in the data. Please check whether this is the case and correct when necessary */ replace eduyears = 0 if age< 10 /*The variable "eduyears" was replaced with a '0' given that the criteria for this indicator is household member aged 10 years or older */ lab var eduyears "Total number of years of education accomplished" /*A control variable is created on whether there is information on years of education for at least 2/3 of the household members aged 10 years and older */ gen temp = 1 if eduyears!=. & age>=10 & age!=. bysort hh_id: egen no_missing_edu = sum(temp) /*Total household members who are 10 years and older with no missing years of education */ gen temp2 = 1 if age>=10 & age!=. bysort hh_id: egen hhs = sum(temp2) /*Total number of household members who are 10 years and older */ replace no_missing_edu = no_missing_edu/hhs replace no_missing_edu = (no_missing_edu>=2/3) /*Identify whether there is information on years of education for at least 2/3 of the household members aged 10 years and older */ tab no_missing_edu, miss label var no_missing_edu "No missing edu for at least 2/3 of the HH members aged 10 years & older" drop temp temp2 hhs /*The entire household is considered deprived if no household member aged 10 years or older has completed SIX years of schooling. */ gen years_edu6 = (eduyears>=6) /* The years of schooling indicator takes a value of "1" if at least someone in the hh has reported 6 years of education or more */ replace years_edu6 = . if eduyears==. bysort hh_id: egen hh_years_edu6_1 = max(years_edu6) gen hh_years_edu6 = (hh_years_edu6_1==1) replace hh_years_edu6 = . if hh_years_edu6_1==. replace hh_years_edu6 = . if hh_years_edu6==0 & no_missing_edu==0 lab var hh_years_edu6 "Household has at least one member with 6 years of edu" ******************************************************************************** *** Step 2.2 Child School Attendance *** ******************************************************************************** codebook ed5, tab (10) gen attendance = . replace attendance = 1 if ed5==1 //Replace attendance with '1' if currently attending school replace attendance = 0 if ed5==2 //Replace attendance with '0' if currently not attending school replace attendance = 0 if ed3==2 //Replace attendance with '0' if never ever attended school tab age ed5, miss //Check individuals who are not of school age replace attendance = 0 if age<5 | age>24 //Replace attendance with '0' for individuals who are not of school age tab attendance, miss label define lab_attend 1 "currently attending" 0 "not currently attending" label values attendance lab_attend label var attendance "Attended school during current school year" /*The entire household is considered deprived if any school-aged child is not attending school up to class 8. */ gen child_schoolage = (age>=6 & age<=14) /* In South Sudan, the official school entrance age is 6 years So, age range is 6-14 (=6+8). Source: "http://data.uis.unesco.org/?ReportId=163" Go to Education>Education>System>Official entrance age to primary education. Look at the starting age and add 8. */ /*A control variable is created on whether there is no information on school attendance for at least 2/3 of the school age children */ count if child_schoolage==1 & attendance==. //Understand how many eligible school aged children are not attending school gen temp = 1 if child_schoolage==1 & attendance!=. bysort hh_id: egen no_missing_atten = sum(temp) /*Total school age children with no missing information on school attendance */ gen temp2 = 1 if child_schoolage==1 bysort hh_id: egen hhs = sum(temp2) //Total number of household members who are of school age replace no_missing_atten = no_missing_atten/hhs replace no_missing_atten = (no_missing_atten>=2/3) /*Identify whether there is missing information on school attendance for more than 2/3 of the school age children */ tab no_missing_atten, miss label var no_missing_atten "No missing school attendance for at least 2/3 of the school aged children" drop temp temp2 hhs bysort hh_id: egen hh_children_schoolage = sum(child_schoolage) replace hh_children_schoolage = (hh_children_schoolage>0) //Control variable: //It takes value 1 if the household has children in school age lab var hh_children_schoolage "Household has children in school age" gen child_not_atten = (attendance==0) if child_schoolage==1 replace child_not_atten = . if attendance==. & child_schoolage==1 bysort hh_id: egen any_child_not_atten = max(child_not_atten) gen hh_child_atten = (any_child_not_atten==0) replace hh_child_atten = . if any_child_not_atten==. replace hh_child_atten = 1 if hh_children_schoolage==0 replace hh_child_atten = . if hh_child_atten==1 & no_missing_atten==0 /*If the household has been intially identified as non-deprived, but has missing school attendance for at least 2/3 of the school aged children, then we replace this household with a value of '.' because there is insufficient information to conclusively conclude that the household is not deprived */ lab var hh_child_atten "Household has all school age children up to class 8 in school" tab hh_child_atten, miss /*Note: The indicator takes value 1 if ALL children in school age are attending school and 0 if there is at least one child not attending. Households with no children receive a value of 1 as non-deprived. The indicator has a missing value only when there are all missing values on children attendance in households that have children in school age. */ ******************************************************************************** *** Step 2.3 Nutrition *** ******************************************************************************** ******************************************************************************** *** Step 2.3a Child Nutrition *** ******************************************************************************** *** Child Underweight Indicator *** ************************************************************************ bysort hh_id: egen temp = max(underweight) gen hh_no_underweight = (temp==0) //Takes value 1 if no child in the hh is underweight replace hh_no_underweight = . if temp==. replace hh_no_underweight = 1 if no_child_eligible==1 //Households with no eligible children will receive a value of 1 lab var hh_no_underweight "Household has no child underweight - 2 stdev" drop temp *** Child Stunting Indicator *** ************************************************************************ bysort hh_id: egen temp = max(stunting) gen hh_no_stunting = (temp==0) //Takes value 1 if no child in the hh is stunted replace hh_no_stunting = . if temp==. replace hh_no_stunting = 1 if no_child_eligible==1 lab var hh_no_stunting "Household has no child stunted - 2 stdev" drop temp *** Child Either Stunted or Underweight Indicator *** ************************************************************************ gen uw_st = 1 if stunting==1 | underweight==1 replace uw_st = 0 if stunting==0 & underweight==0 replace uw_st = . if stunting==. & underweight==. bysort hh_id: egen temp = max(uw_st) gen hh_no_uw_st = (temp==0) //Takes value 1 if no child in the hh is underweight or stunted replace hh_no_uw_st = . if temp==. replace hh_no_uw_st = 1 if no_child_eligible==1 //Households with no eligible children will receive a value of 1 lab var hh_no_uw_st "Household has no child underweight or stunted" drop temp ******************************************************************************** *** Step 2.3b Household Nutrition Indicator *** ******************************************************************************** /* The indicator takes value 1 if there is no children under 5 underweight or stunted. It also takes value 1 for the households that have no eligible children. The indicator takes value missing "." only if all eligible children have missing information in their respective nutrition variable. */ ************************************************************************ gen hh_nutrition_uw_st = 1 replace hh_nutrition_uw_st = 0 if hh_no_uw_st==0 replace hh_nutrition_uw_st = . if hh_no_uw_st==. replace hh_nutrition_uw_st = 1 if no_child_eligible==1 /*We replace households that do not have the applicable population, that is, children 0-5, as non-deprived in nutrition*/ lab var hh_nutrition_uw_st "Household has no child underweight or stunted" ******************************************************************************** *** Step 2.4 Child Mortality *** ******************************************************************************** //NOTE: South Sudan MICS 2010: No information on child mortality from men codebook cm9a cm9b //cm9a: number of sons who have died //cm9b: number of daughters who have died egen temp_f = rowtotal(cm9a cm9b), missing **Total child mortality reported by eligible women replace temp_f = 0 if cm1==1 & cm8==2 | cm1==2 replace temp_f = 0 if cm1==. & cm8==. & marital==1 & temp_f==. /*Assign a value of "0" for: - all eligible women who have ever gave birth but reported no child death - all eligible women who never ever gave birth - all elegible but never married women were not asked birth history questions and hence we assume there is no child mortality among this group*/ replace temp_f = 0 if no_fem_eligible==1 /*Assign a value of "0" for: - individuals living in households that have non-eligible women */ bysort hh_id: egen child_mortality_f = sum(temp_f), missing lab var child_mortality_f "Occurrence of child mortality reported by women" tab child_mortality_f, miss drop temp_f egen child_mortality = rowmax(child_mortality_f) lab var child_mortality "Total child mortality within household reported by women & men" tab child_mortality, miss /*Deprived if any children died in the household in the last 5 years from the survey year */ ************************************************************************ tab child_died_per_wom_5y, miss /* The 'child_died_per_wom_5y' variable was constructed in Step 1.2 using information from individual women who ever gave birth in the BR file. The missing values represent eligible woman who have never ever given birth and so are not present in the BR file. But these 'missing women' may be living in households where there are other women with child mortality information from the BR file. So at this stage, it is important that we aggregate the information that was obtained from the BR file at the household level. This ensures that women who were not present in the BR file is assigned with a value, following the information provided by other women in the household.*/ replace child_died_per_wom_5y = 0 if cm1==2 replace child_died_per_wom_5y = 0 if cm1==. & cm8==. & marital==1 & child_died_per_wom_5y==. /*Assign a value of "0" for: - all eligible women who never ever gave birth - all elegible but never married women were not asked birth history questions and hence we assume there is no child mortality among this group */ replace child_died_per_wom_5y = 0 if no_fem_eligible==1 /*Assign a value of "0" for: - individuals living in households that have non-eligible women */ bysort hh_id: egen child_mortality_5y = sum(child_died_per_wom_5y), missing replace child_mortality_5y = 0 if child_mortality_5y==. & child_mortality==0 /*Replace all households as 0 death if women has missing value and men reported no death in those households */ label var child_mortality_5y "Total child mortality within household past 5 years reported by women" tab child_mortality_5y, miss /* The new standard MPI indicator takes a value of "1" if eligible women within the household reported no child mortality or if any child died longer than 5 years from the survey year. The indicator takes a value of "0" if women in the household reported any child mortality in the last 5 years from the survey year. Households were replaced with a value of "1" if eligible men within the household reported no child mortality in the absence of information from women. The indicator takes a missing value if there was missing information on reported death from eligible individuals. */ gen hh_mortality_5y = (child_mortality_5y==0) replace hh_mortality_5y = . if child_mortality_5y==. lab var hh_mortality_5y "Household had no child mortality in the last 5 years" tab hh_mortality_5y, miss ******************************************************************************** *** Step 2.5 Electricity *** ******************************************************************************** /*Members of the household are considered deprived if the household has no electricity */ clonevar electricity = hc8a codebook electricity, tab (10) replace electricity = 0 if electricity==2 //0=no; 1=yes replace electricity = . if electricity==9 //replace missing values label var electricity "Household has electricity" ******************************************************************************** *** Step 2.6 Sanitation *** ******************************************************************************** /*Members of the household are considered deprived if the household's sanitation facility is not improved, according to MDG guidelines, or it is improved but shared with other household. In cases of mismatch between the MDG guideline and country report, we followed the country report. */ clonevar toilet = ws8 codebook toilet, tab(30) codebook ws9, tab(30) clonevar shared_toilet = ws9 recode shared_toilet (2=0) tab ws9 shared_toilet, miss nol //0=no;1=yes;.=missing /*Note: Following the MDG definition, 'flush to unknown place/not sure/do not know where' is considered as non-improved sanitation facility. However, in South Sudan, this particular category is identified as improved sanitation facilities (p.85 country report)*/ gen toilet_mdg = 1 if (toilet==11 | toilet==12 | toilet==13 | toilet==15 | /// toilet==21 | toilet==22 | toilet==31) & shared_toilet!=1 /*Household is assigned a value of '1' if it uses improved sanitation and does not share toilet with other households */ replace toilet_mdg = 0 if toilet==14 | toilet==23 | toilet==41 | /// toilet==51 | toilet==95 | toilet==96 replace toilet_mdg = 0 if shared_toilet==1 /*Household is assigned a value of '0' if it uses improved sanitation but shares toilet with other households */ replace toilet_mdg = . if toilet==. | toilet==99 //Household is assigned a value of '.' if it has missing information lab var toilet_mdg "Household has improved sanitation with MDG Standards" tab toilet toilet_mdg, miss ******************************************************************************** *** Step 2.7 Drinking Water *** ******************************************************************************** /*Members of the household are considered deprived if the household does not have access to safe drinking water according to MDG guidelines, or safe drinking water is more than a 30-minute walk from home roundtrip. In cases of mismatch between the MDG guideline and country report, we followed the country report.*/ clonevar water = ws1 clonevar timetowater = ws4 codebook water, tab(99) clonevar ndwater = ws2 tab ws2 if water==91 /*Because the quality of bottled water is not known, households using bottled water for drinking are classified as using an improved or unimproved source according to their water source for non-drinking activities such as cooking and hand washing. However, it is important to note that households using bottled water for drinking are classified as unimproved source if this is explicitly mentioned in the country report. */ gen water_mdg = 1 if water==11 | water==12 | water==14 | water==21 | /// water==31 | water==41 | water==51 | water==61 | /// water==91 replace water_mdg = 0 if water==32 | water==42 | water==52 | /// water==62 | water==96 replace water_mdg = 0 if water_mdg==1 & (timetowater>=30 & timetowater<990) & timetowater!=. replace water_mdg = . if water==. | water==99 replace water_mdg = 0 if water==91 & /// (ndwater==32 | ndwater==42 | ndwater==52 | /// ndwater==62 | ndwater==96) /*Households using bottled water for drinking are classified as using an improved or unimproved source according to their water source for non-drinking activities */ lab var water_mdg "Household has drinking water with MDG standards (considering distance)" tab water water_mdg, miss ******************************************************************************** *** Step 2.8 Housing *** ******************************************************************************** /* Members of the household are considered deprived if the household has a dirt, sand or dung floor */ clonevar floor = hc3 codebook floor, tab(99) gen floor_imp = 1 replace floor_imp = 0 if floor<=12 | floor == 96 replace floor_imp = . if floor==99 replace floor_imp = . if floor == . lab var floor_imp "Household has floor that it is not earth/sand/dung" tab floor floor_imp, miss /* Members of the household are considered deprived if the household has wall made of natural or rudimentary materials */ clonevar wall = hc5 codebook wall, tab(99) gen wall_imp = 1 replace wall_imp = 0 if wall<=26| wall==96 replace wall_imp = . if wall==99 replace wall_imp = . if wall == . lab var wall_imp "Household has wall that it is not of low quality materials" tab wall wall_imp, miss /* Members of the household are considered deprived if the household has roof made of natural or rudimentary materials */ clonevar roof = hc4 codebook roof, tab(99) gen roof_imp = 1 replace roof_imp = 0 if roof<=24 | roof==96 replace roof_imp = . if roof==99 replace roof_imp = . if roof== . lab var roof_imp "Household has roof that it is not of low quality materials" tab roof roof_imp, miss /*Household is deprived in housing if the roof, floor OR walls uses low quality materials.*/ gen housing_1 = 1 replace housing_1 = 0 if floor_imp==0 | wall_imp==0 | roof_imp==0 replace housing_1 = . if floor_imp==. & wall_imp==. & roof_imp==. lab var housing_1 "Household has roof, floor & walls that it is not low quality material" tab housing_1, miss ******************************************************************************** *** Step 2.9 Cooking Fuel *** ******************************************************************************** /* Members of the household are considered deprived if the household cooks with solid fuels: wood, charcoal, crop residues or dung. "Indicators for Monitoring the Millennium Development Goals", p. 63 */ clonevar cookingfuel = hc6 codebook cookingfuel, tab(99) gen cooking_mdg = 1 replace cooking_mdg = 0 if cookingfuel>5 & cookingfuel<95 replace cooking_mdg = . if cookingfuel==. |cookingfuel==99 lab var cooking_mdg "Household has cooking fuel according to MDG standards" /* Deprived if: 6 "coal/lignite", 7 "charcoal", 8 "wood", 9 "straw/shrubs/grass" 10 "agricultural crop", 11 "animal dung" */ tab cookingfuel cooking_mdg, miss /*Note that in South Sudan, the category 'other' cooking fuel is not identified as solid fuel in the country report. Hence this particular category is identified as 'non-deprived' */ ******************************************************************************** *** Step 2.10 Assets ownership *** ******************************************************************************** /* Members of the household are considered deprived if the household does not own more than one of: radio, TV, telephone, bike, motorbike or refrigerator and does not own a car or truck. */ clonevar television = hc8c gen bw_television = . clonevar radio = hc8b clonevar telephone = hc8d clonevar mobiletelephone = hc9b clonevar refrigerator = hc8e clonevar car = hc9f clonevar bicycle = hc9c clonevar motorbike = hc9d clonevar computer = hc9f clonevar internet = hc8g clonevar motorboat = hc9g clonevar animal_cart = hc9e clonevar land = hc11 foreach var in television radio telephone mobiletelephone refrigerator /// car bicycle motorbike computer animal_cart { replace `var' = 0 if `var'==2 replace `var' = . if `var'==9 | `var'==99 | `var'==8 | `var'==98 } //Missing values replaced //Group telephone and mobiletelephone as a single variable replace telephone=1 if telephone==0 & mobiletelephone==1 replace telephone=1 if telephone==. & mobiletelephone==1 /* Members of the household are considered deprived in assets if the household does not own more than one of: radio, TV, telephone, bike, motorbike, refrigerator, computer or animal_cart and does not own a car or truck.*/ egen n_small_assets2 = rowtotal(television radio telephone refrigerator bicycle motorbike computer animal_cart), missing lab var n_small_assets2 "Household Number of Small Assets Owned" gen hh_assets2 = (car==1 | n_small_assets2 > 1) replace hh_assets2 = . if car==. & n_small_assets2==. lab var hh_assets2 "Household Asset Ownership: HH has car or more than 1 small assets incl computer & animal cart" ******************************************************************************** *** Step 2.11 Rename and keep variables for MPI calculation ******************************************************************************** //Retain data on sampling design: desc psu stratum rename stratum strata //Retain year, month & date of interview: desc hh5y hh5m hh5d clonevar year_interview = hh5y clonevar month_interview = hh5m clonevar date_interview = hh5d //Generate presence of subsample gen subsample = . *** Rename key global MPI indicators for estimation *** recode hh_mortality_5y (0=1)(1=0) , gen(d_cm) recode hh_nutrition_uw_st (0=1)(1=0) , gen(d_nutr) recode hh_child_atten (0=1)(1=0) , gen(d_satt) recode hh_years_edu6 (0=1)(1=0) , gen(d_educ) recode electricity (0=1)(1=0) , gen(d_elct) recode water_mdg (0=1)(1=0) , gen(d_wtr) recode toilet_mdg (0=1)(1=0) , gen(d_sani) recode housing_1 (0=1)(1=0) , gen(d_hsg) recode cooking_mdg (0=1)(1=0) , gen(d_ckfl) recode hh_assets2 (0=1)(1=0) , gen(d_asst) *** Keep selected variables for global MPI estimation *** keep hh_id ind_id ccty ccnum cty survey year subsample /// strata psu weight area relationship sex age agec7 agec4 marital hhsize /// region year_interview month_interview date_interview /// d_cm d_nutr d_satt d_educ d_elct d_wtr d_sani d_hsg d_ckfl d_asst order hh_id ind_id ccty ccnum cty survey year subsample /// strata psu weight area relationship sex age agec7 agec4 marital hhsize /// region year_interview month_interview date_interview /// d_cm d_nutr d_satt d_educ d_elct d_wtr d_sani d_hsg d_ckfl d_asst *** Sort, compress and save data for estimation *** sort ind_id compress save "$path_out/ssd_mics10_pov.dta", replace log close /*Concern: Some 37% of children under 5 have either missing info on height or weight. The country report confirmed the high missing anthropometric information (see page 27 of the country report). The report further stated that the unusual high missing values may be because of the failure of enumerators to patiently collect the required information from children under 5 (p.9). As a result, this affect the size of missing observations for the final nutrition indicator. Some 23 percent of the individuals live in households where there is missing nutrition data. The report also indicated high non-response rate among women aged 15-49 years. This resulted in high missing values for the question on child mortality and birth history of women 15-49 years (p.8). */