clear
cap log close
cap clear matrix
set more off
set mem 200m
set matsize 800
program drop _all
version 11.0

gl root 		"C:\DROPBOX\Dropbox\TESTS_COLOMBIA"

gl testdata 	"$root\DATA\CLEANDATA\v4_tests\idb_portal\tests_bogota.dta"
gl out 			"$root\DATA\CODE\v4_analysis_tests\idb_portal\descriptives"

log using "$out/sample charact.log", replace

use "$testdata"

********CHILD CHARACTERISTICS Variables********
*12-mth of age intervals
gen m6t18=0
	replace m6t18=1 if b_age_mth >= 5 & b_age_mth <= 18 
	
gen m19t30=0
	replace m19t30=1 if b_age_mth >= 19 & b_age_mth <= 30 
	
gen m31t42=0
	replace m31t42=1 if b_age_mth >= 31 & b_age_mth <= 42 
	
	*6-mth of age intervals
gen m6t12=0
	replace m6t12=1 if b_age_mth >= 5 & b_age_mth <= 12 
	
gen m13t18=0
	replace m13t18=1 if b_age_mth >= 13 & b_age_mth <= 18 

gen m19t24=0
	replace m19t24=1 if b_age_mth >= 19 & b_age_mth <= 24 
	
gen m25t30=0
	replace m25t30=1 if b_age_mth >= 25 & b_age_mth <= 30 

gen m31t36=0
	replace m31t36=1 if b_age_mth >= 31 & b_age_mth <= 36 
	
gen m37t42=0
	replace m37t42=1 if b_age_mth >= 37 & b_age_mth <= 42 

********PARENTAL CHARACTERISTICS Variables********	
gen more_sec_ed_mo = 0
	replace more_sec_ed_mo = 1 if edu_cat_mo > 3
	replace more_sec_ed_mo =.  if edu_cat_mo ==.
	
gen more_sec_ed_fa = 0
	replace more_sec_ed_fa = 1 if edu_cat_fa > 3
	replace more_sec_ed_fa = . if edu_cat_fa ==.
	
gen no_present_fa = 0
	replace no_present_fa = 1 if dead_fa == 1 | absent_fa == 1
	replace no_present_fa =.  if dead_fa == . & absent_fa == .

********WEALTH DISTRIBUTION Variables**********
gen quartil1 = 0
	replace quartil1 = 1 if wealth_5p_qr == 1
gen quartil2 = 0
	replace quartil2 = 1 if wealth_5p_qr == 2
gen quartil3 = 0
	replace quartil3 = 1 if wealth_5p_qr == 3
gen quartil4 = 0
	replace quartil4 = 1 if wealth_5p_qr == 4
	
gen estrato1 = 0
	replace estrato1 = 1 if estrato == 1
gen estrato2 = 0
	replace estrato2 = 1 if estrato == 2
gen estrato3 = 0
	replace estrato3 = 1 if estrato == 3
gen estrato4 = 0
	replace estrato4 = 1 if estrato == 4
	
	
********FAMILY CARE INDEX Variables**********
gen fci_books_news_score = fci_newspapers + fci_books_adults

	
********CHILD CHARACTERISTICS Labels********	
label var m6t18  "6 to 18 months = 1"
label var m19t30 "19 to 30 months = 1"
label var m31t42 "31 to 42 months = 1"
label var m6t12  "6 to 12 months = 1"
label var m13t18 "13 to 18 months = 1"
label var m19t24 "19 to 24 months = 1"
label var m25t30 "25 to 30 months = 1"
label var m31t36 "31 to 36 months = 1"
label var m37t42 "37 to 42 months = 1"

label var female "Female = 1"
label var premature "Premature (gestional age <37 weeks) = 1"
label var birthweight "Birth weight in gr."
label var first_child "First born = 1"

********PARENTAL CHARACTERISTICS Labels********
label var age_mo "Age Mother"
label var edu_mo "Education Years Mother"
label var more_sec_ed_mo "Mother has more than Secondary Education = 1"
label var work_mo "Mother Works (paid or unpaid) = 1"
label var young_birth_mo "Mother Gave Birth Before Age 18 = 1"
label var edu_fa "Education Years Father"
label var more_sec_ed_fa "Father has more than Secondary Education = 1"
label var no_present_fa "Father Deceased or No Longer Living with Child = 1"

********HOUSEHOLD CHARACTERISTICS Labels**********
label var hhsize "Household Size"
label var grandma_mo "Grandmother Lives in Houshold = 1"
label var crowding "Crowding (people per room)"
label var floors_quality "Quality Floors (tiles carpet wood) = 1"
label var windows "External Windows = 1"
label var kitchen_shared "Shared Kitchen = 1"
label var bathroom_shared "Shared Bathroom = 1"
label var bathroom_dnb "More than One Bathroom = 1"
label var car "Car = 1"
label var fridge "Fridge = 1"
label var microwave "Microwave = 1"
label var washer "Washing Machine = 1"
label var boiler "Boiler = 1"
label var computer "Computer = 1"
label var smartphone "Smartphone = 1"
label var flat_tv "Flat TV = 1"
label var home_theatre "Home Theatre = 1"
label var dvd "DVD = 1"
label var stereo "Stereo = 1"
label var consola "Games Console = 1"
label var internet "Internet = 1"

********WEALTH DISTRIBUTION*******
label var wealth_5p "Wealth Index"
label var quartil1 "Quartil 1 = 1"
label var quartil2 "Quartil 2 = 1"
label var quartil3 "Quartil 3 = 1"
label var quartil4 "Quartil 4 = 1"
label var estrato1 "Estrato 1 = 1"
label var estrato2 "Estrato 2 = 1"
label var estrato3 "Estrato 3 = 1"
label var estrato4 "Estrato 4 = 1"

*******HOME STIMULATION Labels********
label var fci_books_news_score "Books and Newspapers (FCI Score)"
label var fci_play_materials_1 "Play Materials (FCI Score)"
label var fci_play_activities_1 "Play Activities (FCI Score)"


*******CHILD CARE ARRENGMENTS Labels********
label var care_center "Child Care Center Attendance = 1"
label var care_minder "Care Minder = 1"


*******CATEGORIES**************
local child_vars	m6t18 m19t30 m31t42 m6t12 m13t18 m19t24 m25t30 m31t36 m37t42 female premature birthweight first_child stunted
local par_vars		age_mo edu_mo more_sec_ed_mo work_mo young_birth_mo edu_fa more_sec_ed_fa no_present_fa
local hh_vars		hhsize crowding grandma_mo floors_quality windows kitchen_shared bathroom_shared bathroom_dnb ///
					car fridge microwave washer boiler computer smartphone flat_tv home_theatre dvd stereo consola internet
local fci_vars		fci_books_news_score fci_play_materials_1 fci_play_activities_1
local care_vars		care_center care_minder
local wealth		wealth_5p quartil1 quartil2 quartil3 quartil4 estrato1 estrato2 estrato3 estrato4

********SUMMARIZE fo Bayley-III subsample!********	
estpost summarize 	`child_vars' `par_vars' `hh_vars' `wealth' `fci_vars' `care_vars'  ///
					if battery==2 // Battery A (ASQ3, Denver, MacArthur)
est store A

estpost summarize 	`child_vars' `par_vars' `hh_vars' `wealth' `fci_vars' `care_vars' ///
					if battery==1 // Battery B (Battelle, Adp. Behav., Milestones)
est store B

estpost summarize 	`child_vars' `par_vars' `hh_vars' `wealth' `fci_vars' `care_vars' 
est store C

estpost ttest	 	`child_vars' `par_vars' `hh_vars' `wealth' `fci_vars' `care_vars', by(battery)
est store D

********EXPORT********
	esttab C A B D using "$out\char_per_battery.csv", replace  wide obslast plain ///
	title(Table 1: Mean Sample Characteristics by Battery) ///
	refcat(m6t18 "I. Child Characteristics" age_mo "II. Parental Characteristics" hhsize "III. Household Characteristics"  ///
	wealth_5p "IV. Wealth Distribution" fci_books_news_score "V. Level Home Stimulation" care_center "VI. Child Care Arrangements", nolabel) ///
	collabel("N" "Mean" "SD" "Mean dif." "p-value") ///
	mtitle("Total" "Battery A" "Battery B" "Battery A - Battery B") ///
	cells ("count(fmt(3)) mean(fmt(3)) sd(fmt(3)) b(fmt(3)) p(fmt(3))") label lines nonum noobs ///
	addnote("Data are means. SD reported.") ///
	stats(N, fmt(%18.0g) labels("Observations"))

	
********** OBTAINING p-values of DIFFERENCES BY CLUSTERS (seccion) **********
#delimit;	
replace battery = 0 if battery == 2;

gen varname = "";
gen meanall =.;
gen sdall = .;
gen obsall =.;
gen obsbat0 =.;
gen meanbat0 = . ;
gen sdbat0 = . ;
gen obstat1 = .;
gen meanbat1 = .;
gen sdbat1 = .;
gen diffmean =.;
gen pval =.;

local j = 1;

	foreach var in `child_vars' {;
	replace varname="`var'" in `j' ;
		foreach y in 0 1 {;
			sum `var' if battery == `y';
			replace meanbat`y' =r(mean) in `j';
			replace sdbat`y' =r(sd) in `j';
			scalar obstat`y' =r(N) in `j';
			};
			sum `var';
			replace meanall =r(mean) in `j';
			replace sdall =r(sd) in `j';
			scalar obsall`y' =r(N) in `j';
		
		reg `var' battery, cl(secc);
		replace diffmean = _b[battery] in `j';
		local t = _b[battery]/_se[battery];
		replace pval = 2*ttail(e(df_r),abs(`t')) in `j';
		
		local j = `j'+1;
		};
		
		local j = `j' + 1;
		
	foreach var in  `par_vars' {;
	replace varname="`var'" in `j' ;
		foreach y in 0 1 {;
			sum `var' if battery == `y';
			replace meanbat`y' =r(mean) in `j';
			replace sdbat`y' =r(sd) in `j';
			scalar obstat`y' =r(N) in `j';
			};
			sum `var';
			replace meanall =r(mean) in `j';
			replace sdall =r(sd) in `j';
			scalar obsall`y' =r(N) in `j';
		
		reg `var' battery, cl(secc);
		replace diffmean = _b[battery] in `j';
		local t = _b[battery]/_se[battery];
		replace pval = 2*ttail(e(df_r),abs(`t')) in `j';
		
		local j = `j'+1;
		};		

		local j = `j' + 1;
		
	foreach var in `hh_vars' {;
	replace varname="`var'" in `j' ;
		foreach y in 0 1 {;
			sum `var' if battery == `y';
			replace meanbat`y' =r(mean) in `j';
			replace sdbat`y' =r(sd) in `j';
			scalar obstat`y' =r(N) in `j';
			};
			sum `var';
			replace meanall =r(mean) in `j';
			replace sdall =r(sd) in `j';
			scalar obsall`y' =r(N) in `j';
			
		reg `var' battery, cl(secc);
		replace diffmean = _b[battery] in `j';
		local t = _b[battery]/_se[battery];
		replace pval = 2*ttail(e(df_r),abs(`t')) in `j';
		
		local j = `j'+1;
		};
		
		local j = `j' + 1;
		
	foreach var in `wealth' {;
	replace varname="`var'" in `j' ;
		foreach y in 0 1 {;
			sum `var' if battery == `y';
			replace meanbat`y' =r(mean) in `j';
			replace sdbat`y' =r(sd) in `j';
			scalar obstat`y' =r(N) in `j';
			};
			sum `var';
			replace meanall =r(mean) in `j';
			replace sdall =r(sd) in `j';
			scalar obsall`y' =r(N) in `j';
			
		reg `var' battery, cl(secc);
		replace diffmean = _b[battery] in `j';
		local t = _b[battery]/_se[battery];
		replace pval = 2*ttail(e(df_r),abs(`t')) in `j';
		
		local j = `j'+1;
		};
		
		local j = `j' + 1;	
		
	foreach var in `fci_vars' {;
	replace varname="`var'" in `j' ;
		foreach y in 0 1 {;
			sum `var' if battery == `y';
			replace meanbat`y' =r(mean) in `j';
			replace sdbat`y' =r(sd) in `j';
			scalar obstat`y' =r(N) in `j';
			};
			sum `var';
			replace meanall =r(mean) in `j';
			replace sdall =r(sd) in `j';
			scalar obsall`y' =r(N) in `j';	
			
		reg `var' battery, cl(secc);
		replace diffmean = _b[battery] in `j';
		local t = _b[battery]/_se[battery];
		replace pval = 2*ttail(e(df_r),abs(`t')) in `j';
		
		local j = `j'+1;
		};	
		
		local j = `j' + 1;
		
	foreach var in `care_vars' {;
	replace varname="`var'" in `j' ;
		foreach y in 0 1 {;
			sum `var' if battery == `y';
			replace meanbat`y' =r(mean) in `j';
			replace sdbat`y' =r(sd) in `j';
			scalar obstat`y' =r(N) in `j';
			};
			sum `var';
			replace meanall =r(mean) in `j';
			replace sdall =r(sd) in `j';
			scalar obsall`y' =r(N) in `j';
			
		reg `var' battery, cl(secc);
		replace diffmean = _b[battery] in `j';
		local t = _b[battery]/_se[battery];
		replace pval = 2*ttail(e(df_r),abs(`t')) in `j';
		
		local j = `j'+1;
		};
		
		local j = `j' + 1;
		

keep varname meanall sdall meanbat0 sdbat0 meanbat1 sdbat1 diffmean pval;
save "$out/sample charact_ttests.dta", replace;

log close;