Revised refactored regresssions

input/InitialPopulations/compile/RegressionEstimates/00_master_regression_estimates.do

@@ -72,7 +72,7 @@ global estimation_sample "${path}\initial_populations\data\ukhls_pooled_ipop.dta
 * Pooled dataset with predicted wages after Heckman   
 global estimation_sample2 "${path}\initial_populations\data\UKHLS_pooled_ipop2.dta"
 
-* Directory containing external data used for the estimates (e.g. fertility rates, wage growth) 
+* Directory containing external data used +for the estimates (e.g. fertility rates, wage growth) 
 global dir_external_data "${dir_work}/external_data"
 
 * Directory to save data for internal validation 
@@ -92,7 +92,7 @@ do "${path}\00_master_conditions.do"
 Two additional do-files are called from each of these do-files
 - variable_update.do refactors variable names 
 - programs.do contains Stata programs to process the output of regressions and create Excel files with results used by Simpaths 
- */
+*/
 
 do "${dir_do}/01_reg_education.do"
 
@@ -114,7 +114,9 @@ do "${dir_do}/09_reg_income.do"
 
 do "${dir_do}/10_reg_socialcare.do" 
 
+
 /*Note that the do-files below are not yet refactored  */
+
 do "${dir_do}/11_reg_financial_distress.do"
 
 do "${dir_do}/12_reg_health_mental.do"

input/InitialPopulations/compile/RegressionEstimates/01_reg_education.do

@@ -26,7 +26,7 @@ log using "${dir_log}/reg_education.log", replace
 
 /******************************* SET EXCEL FILE *******************************/
 
-putexcel set "$dir_results/reg_education", sheet("Info") replace
+putexcel set "$dir_results/reg_education", sheet("Info") modify //replace
 putexcel A1 = "Description:", bold
 putexcel B1 = "Model parameters governing projection of education status"
 putexcel A2 = "Authors:"
@@ -79,34 +79,55 @@ do "${dir_do}/programs.do"
 /****************** E1a: PROBABILITY OF REMAINING IN EDUCATION ****************/
 display "${e1a_if_condition}"	
 
-
+/*
+probit Dst i.Dgn Dag Dag_sq /*Dag_c Dag_c_sq Dag_post18_sq*/  li.Ded  ///
+	li.Dehmf_c3_Medium li.Dehmf_c3_Low ///
+	li.Ydses_c5_Q2 li.Ydses_c5_Q3 li.Ydses_c5_Q4 li.Ydses_c5_Q5 ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	if ${e1a_if_condition} [pw=${weight}], vce(robust)	
+
+process_regression, domain("education") process("E1a") sheet("E1a_orig") ///
+	title("Process E1a: Prob. remain in education") ///
+	gofrow(3) goflabel("E1a - Remain in education") ///
+	ifcond("${e1a_if_condition}") probit	
+	*/
 probit Dst ///
     demMaleFlag demAge demAgeSq eduSampleFlagL1  ///
 	eduHighestParentC3MediumL1 eduHighestParentC3LowL1  ///
 	yHhQuintilesMonthC5Q2L1 yHhQuintilesMonthC5Q3L1 yHhQuintilesMonthC5Q4L1 yHhQuintilesMonthC5Q5L1 ///
-	$regions demYear demYear2020 demYear2021 $ethnicity ///
+	$regions demYearTransformed demYear2020 demYear2021 $ethnicity ///
 	if ${e1a_if_condition} [pw=${weight}], vce(robust)	
 
-
 process_regression, domain("education") process("E1a") sheet("E1a") ///
 	title("Process E1a: Prob. remain in education") ///
 	gofrow(3) goflabel("E1a - Remain in education") ///
 	ifcond("${e1a_if_condition}") probit	 	
 
-
 
 /****************** E1b: PROBABILITY OF RETURNING TO EDUCATION ****************/
 display "${e1b_if_condition}"	
 
+/*
+probit der i.Dgn Dag Dag_sq li.Dcpst_Partnered ///
+	li.Deh_c4_High li.Deh_c4_Low li.Dehmf_c3_Medium li.Dehmf_c3_Low ///
+	li.Les_c3_NotEmployed li.Les_c3_Employed l.Dnc l.Dnc02 ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	if ${e1b_if_condition} [pw=${weight}], vce(robust)
+
+process_regression, domain("education") process("E1b") sheet("E1b_orig") ///
+	title("Process E1b: Prob. return to education") ///
+	gofrow(7) goflabel("E1b - Return to education") ///
+	ifcond("${e1b_if_condition}") probit	
+	*/
 probit der ///
     demMaleFlag demAge demAgeSq  demPartnerStatusPartneredL1 ///
 	eduHighestC4HighL1 eduHighestC4LowL1 ///
 	eduHighestParentC3MediumL1 eduHighestParentC3LowL1 ///
-	labStatusC3NotEmployedL1 /*labStatusC3EmployedL1*/ ///
+	labStatusC3NotEmployedL1 labStatusC3EmployedL1 ///
 	demNChildL1 demNChild0to2L1 ///
-	$regions demYear demYear2020 demYear2021 $ethnicity ///
+	$regions demYearTransformed demYear2020 demYear2021 $ethnicity ///
 	if ${e1b_if_condition} [pw=${weight}], vce(robust)
-
+	
 process_regression, domain("education") process("E1b") sheet("E1b") ///
 	title("Process E1b: Prob. return to education") ///
 	gofrow(7) goflabel("E1b - Return to education") ///
@@ -116,13 +137,24 @@ process_regression, domain("education") process("E1b") sheet("E1b") ///
 /****************** E2: EDUCATION ATTAINMENT WHEN LEAVE SCHOOL ****************/
 display "${e2_if_condition}"	
 
+/*
+gologit2 deh_c3_recoded i.Dgn Dag Dag_sq ///
+	i.L_Dehmf_c3_Medium i.L_Dehmf_c3_Low ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	if ${e2_if_condition} [pw=${weight}], autofit 
+
+process_gologit, domain("education") process("E2") sheet("E2_orig") ///
+    title("Process E2: Educational Attainment When Leave School") ///
+    gofrow(11) goflabel("E2 - Education attainment") ///
+    outcomes(3) ///
+    ifcond("${e2_if_condition}")
+	*/
 gologit2 deh_c3_recoded ///
     demMaleFlag demAge demAgeSq ///
 	eduHighestParentC3MediumL1 eduHighestParentC3LowL1 ///
-	$regions demYear demYear2020 demYear2021 $ethnicity ///
+	$regions demYearTransformed demYear2020 demYear2021 $ethnicity ///
 	if ${e2_if_condition} [pw=${weight}] , autofit 
 
-
 process_gologit, domain("education") process("E2") sheet("E2") ///
     title("Process E2: Educational Attainment When Leave School") ///
     gofrow(11) goflabel("E2 - Education attainment") ///
@@ -132,5 +164,5 @@ process_gologit, domain("education") process("E2") sheet("E2") ///
 
 display "Education analysis complete!"
 
-
+		 
 capture log close

input/InitialPopulations/compile/RegressionEstimates/02_reg_leave_parental_home.do

@@ -24,7 +24,7 @@ log using "${dir_log}/reg_leave_parental_home.log", replace
 
 /********************************* SET EXCEL FILE *****************************/
 
-putexcel set "$dir_results/reg_leave_parental_home", sheet("Info") replace
+putexcel set "$dir_results/reg_leave_parental_home", sheet("Info") modify //replace
 putexcel A1 = "Description:", bold
 putexcel B1 = "Model parameters governing leaving parental home"
 putexcel A2 = "Authors:"
@@ -65,13 +65,26 @@ do "${dir_do}/programs.do"
 
 /**************** P1: PROBABILITY OF LEAVING THE PARENTAL HOME ****************/
 display "${p1_if_condition}"
+
+/*
+probit dlftphm i.Dgn Dag Dag_sq li.Deh_c4_Na li.Deh_c4_Medium li.Deh_c4_Low ///
+	li.Les_c3_Student li.Les_c3_NotEmployed ///
+	li.Ydses_c5_Q2 li.Ydses_c5_Q3 li.Ydses_c5_Q4 li.Ydses_c5_Q5 ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	if ${p1_if_condition} [pw=${weight}], vce(robust)
+
+process_regression, domain("leave_parental_home") process("P1") sheet("P1_orig") ///
+	title("Process P1: Prob. leave parental home") ///
+	gofrow(3) goflabel("P1 - Leave parental home") ///
+	ifcond("${p1_if_condition}") probit	
+	*/
 
 probit dlftphm ///
     demMaleFlag demAge demAgeSq  ///
     eduHighestC4NaL1 eduHighestC4MediumL1 eduHighestC4LowL1 ///
 	labStatusC3StudentL1 labStatusC3NotEmployedL1 ///
 	yHhQuintilesMonthC5Q2L1 yHhQuintilesMonthC5Q3L1 yHhQuintilesMonthC5Q4L1 yHhQuintilesMonthC5Q5L1 ///
-	$regions demYear demYear2020 demYear2021 $ethnicity ///
+	$regions demYearTransformed demYear2020 demYear2021 $ethnicity ///
 	if ${p1_if_condition} [pw=${weight}], vce(robust)
 
 

input/InitialPopulations/compile/RegressionEstimates/03_reg_partnership.do

@@ -24,7 +24,7 @@ log using "${dir_log}/reg_partnership.log", replace
 
 /********************************* SET EXCEL FILE *****************************/
 
-putexcel set "$dir_results/reg_partnership", sheet("Info") replace
+putexcel set "$dir_results/reg_partnership", sheet("Info") modify //replace
 putexcel A1 = "Description:", bold
 putexcel B1 = "Model parameters for relationship status projection"
 putexcel A2 = "Authors:"
@@ -66,18 +66,36 @@ do "${dir_do}/programs.do"
 
 /******************** U1: PROBABILITY FORMING PARTNERSHIP *********************/
 display "${u1_if_condition}"	
-
+/*
+probit dcpen ///
+    i.Ded Dgn Dag Dag_sq ///
+    lc.Dnc lc.Dnc02 ///
+    li.Ydses_c5_Q2 li.Ydses_c5_Q3 li.Ydses_c5_Q4 li.Ydses_c5_Q5 ///
+	/*Ded_Dag Ded_Dag_sq*/ Ded_Dgn Ded_Dnc_L1 Ded_Dnc02_L1 ///
+	Ded_Ydses_c5_Q2_L1 Ded_Ydses_c5_Q3_L1 Ded_Ydses_c5_Q4_L1 Ded_Ydses_c5_Q5_L1 ///
+	i.Deh_c4_Na i.Deh_c4_High i.Deh_c4_Medium i.Deh_c4_Low ///
+	li.Les_c4_Student li.Les_c4_NotEmployed li.Les_c4_Retired /// 
+	li.Les_c4_Student_Dgn li.Les_c4_NotEmployed_Dgn li.Les_c4_Retired_Dgn ///
+	l.Dhe_pcs l.Dhe_mcs ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	if ${u1_if_condition} [pw=${weight}], vce(robust)
+
+process_regression, domain("partnership") process("U1") sheet("U1_orig") ///
+	title("Process U1: Prob. form partnership") ///
+	gofrow(3) goflabel("U1 - Form partnership") ///
+	ifcond("${u1_if_condition}") probit	
+*/
 probit dcpen ///
        eduSampleFlag demMaleFlag demAge demAgeSq ///
        demNChildL1 demNChild0to2L1 ///
        yHhQuintilesMonthC5Q2L1 yHhQuintilesMonthC5Q3L1 yHhQuintilesMonthC5Q4L1 yHhQuintilesMonthC5Q5L1 ///
        eduSampleFlag_demMaleFlag eduSampleFlag_demNChildL1 eduSampleFlag_demNChild0to2L1 ///
 	   eduSampleFlag_Q2L1   eduSampleFlag_Q3L1  eduSampleFlag_Q4L1  eduSampleFlag_Q5L1 ///
-	   eduHighestC4NaL1 eduHighestC4HighL1 eduHighestC4MediumL1 eduHighestC4LowL1 ///
-	   labStatusC4EmployedL1 labStatusC4StudentL1 labStatusC4RetiredL1 /// 
-       labStatusC4EmployedL1_Male labStatusC4StudentL1_Male labStatusC4RetiredL1_Male ///
+	   eduHighestC4High eduHighestC4Medium /*eduHighestC4Low*/ ///
+	   labStatusC4StudentL1 labStatusC4NotEmployedL1 labStatusC4RetiredL1 /// 
+       labStatusC4Student_MaleL1 labStatusC4NotEmployed_MaleL1 labStatusC4Retired_MaleL1 ///
        healthPhysicalPcsL1 healthMentalMcsL1  ///
-	   $regions demYear demYear2020 demYear2021 $ethnicity /// 
+	   $regions demYearTransformed demYear2020 demYear2021 $ethnicity /// 
     if ${u1_if_condition} [pw=${weight}], vce(robust)
 
 process_regression, domain("partnership") process("U1") sheet("U1") ///
@@ -88,18 +106,40 @@ process_regression, domain("partnership") process("U1") sheet("U1") ///
 
 /******************* U2: PROBABILITY TERMINATE PARTNERSHIP ********************/
 display "${u2_if_condition}"	
-
+
+/*
+probit dcpex ///
+    i.Ded Dag Dag_sq /*Ded_Dag Ded_Dag_sq*/ ///
+    li.Deh_c4_Na li.Deh_c4_Low  li.Deh_c4_Medium li.Deh_c4_High ///
+	li.Dehsp_c3_Medium li.Dehsp_c3_Low ///
+	li.Dhe_Fair li.Dhe_Good li.Dhe_VeryGood li.Dhe_Excellent  ///
+	l.Dhe_pcs l.Dhe_mcs ///
+	l.Dhe_pcssp l.Dhe_mcssp ///
+	l.Dcpyy l.New_rel l.Dcpagdf ///
+	l.Dnc l.Dnc02 ///
+	li.Lesdf_c4_EmpSpouseNotEmp li.Lesdf_c4_NotEmpSpouseEmp li.Lesdf_c4_BothNotEmployed ///
+	l.Ypnbihs_dv l.Ynbcpdf_dv ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	if ${u2_if_condition} [pw=${weight}], vce(robust)
+
+process_regression, domain("partnership") process("U2") sheet("U2_orig") ///
+	title("Process U2: Prob. end partnership") ///
+	gofrow(7) goflabel("U2 - End partnership") ///
+	ifcond("${u2_if_condition}") probit	
+*/
+
 probit dcpex ///
-      eduSampleFlag demMaleFlag demAge demAgeSq ///
-      eduHighestC4NaL1 eduHighestC4HighL1 eduHighestC4MediumL1 eduHighestC4LowL1 ///
+      eduSampleFlag demAge demAgeSq ///
+      eduHighestC4NaL1 eduHighestC4LowL1 eduHighestC4MediumL1 eduHighestC4HighL1 ///
       eduHighestPartnerC3MediumL1 eduHighestPartnerC3LowL1 ///
-      healthPhysicalPcsL1 healthMentalMcsL1  ///
+      healthSelfRatedFairL1 healthSelfRatedGoodL1 healthSelfRatedVeryGoodL1 healthSelfRatedExcellentL1 ///
+	  healthPhysicalPcsL1 healthMentalMcsL1  ///
 	  healthPhysicalPartnerPcsL1 healthMentalPartnerMcsL1 ///
 	  demPartnerNYearL1 demEnterPartnerFlagL1 demAgePartnerDiffL1 ///
 	  demNChildL1 demNChild0to2L1 ///
 	  labStatusPartnerAndOwnC42L1 labStatusPartnerAndOwnC43L1 labStatusPartnerAndOwnC44L1 ///
 	  yNonBenPersGrossMonthL1 yPersAndPartnerGrossDiffMonthL1 ///
-	  $regions demYear demYear2020 demYear2021 $ethnicity /// 
+	  $regions demYearTransformed demYear2020 demYear2021 $ethnicity /// 
 	if ${u2_if_condition} [pw=${weight}], vce(robust)
 
 process_regression, domain("partnership") process("U2") sheet("U2") ///

input/InitialPopulations/compile/RegressionEstimates/04_reg_fertility.do

@@ -24,7 +24,7 @@ log using "${dir_log}/reg_fertility.log", replace
 
 /******************************* SET EXCEL FILE *******************************/
 
-putexcel set "$dir_results/reg_fertility", sheet("Info") replace
+putexcel set "$dir_results/reg_fertility", sheet("Info") modify //replace
 putexcel A1 = "Description:", bold
 putexcel B1 = "Model parameters governing projection of fertility"
 putexcel A2 = "Authors:"
@@ -69,18 +69,38 @@ do "${dir_do}/programs.do"
 
 /*********************** F1: PROBABILITY OF HAVING A CHILD ********************/
 display "${f1_if_condition}" 
+/*
+probit dchpd ///
+	i.Ded Dag Dag_sq ///
+	l.Dhe_pcs l.Dhe_mcs ///
+	Dcpst_Single li.Dcpst_Single ///
+	/*Ded_Dag Ded_Dhe_pcs Ded_Dhe_mcs*/ ///
+	Ded_Dcpst_Single /*Ded_Dcpst_Single_L1*/ ///
+	li.Ydses_c5_Q2 li.Ydses_c5_Q3 li.Ydses_c5_Q4 li.Ydses_c5_Q5 ///
+	l.Dnc l.Dnc02 ///
+	i.Deh_c4_Low i.Deh_c4_High ///
+	FertilityRate ///
+	/*li.Les_c3_Student*/ li.Les_c3_NotEmployed ///
+	$regions Year_transformed Y2020 Y2021 $ethnicity ///
+	 if ${f1_if_condition} [pw=${weight}], vce(robust)
+
+process_regression, domain("fertility") process("F1") sheet("F1_orig") ///
+	title("Process F1: Prob. have a child") ///
+	gofrow(3) goflabel("F1 - Have child") ///
+	ifcond("${f1_if_condition}") probit	 
+*/
 
 probit dchpd ///
-	eduSampleFlag demMaleFlag demAge demAgeSq ///
+	eduSampleFlag demAge demAgeSq ///
 	healthPhysicalPcsL1 healthMentalMcsL1 ///
 	demPartnerStatusSingle demPartnerStatusSingleL1 ///
 	eduSampleFlag_Single ///
 	yHhQuintilesMonthC5Q2L1 yHhQuintilesMonthC5Q3L1 yHhQuintilesMonthC5Q4L1 yHhQuintilesMonthC5Q5L1 ///
 	demNChildL1 demNChild0to2L1 ///
-	eduHighestC4HighL1 eduHighestC4MediumL1 eduHighestC4LowL1 ///
+	eduHighestC4Low eduHighestC4High ///
 	fertilityRate ///
 	/*labStatusC3StudentL1*/ labStatusC3NotEmployedL1 ///
-	$regions demYear demYear2020 demYear2021 $ethnicity /// 
+	$regions demYearTransformed demYear2020 demYear2021 $ethnicity /// 
 if ${f1_if_condition} [pw=${weight}], vce(robust)
 
 process_regression, domain("fertility") process("F1") sheet("F1") ///