* Encoding: UTF-8. * --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Ethiopia (Addis Ababa + Somali) PPS sampling. * --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- file handle SampPath /name "D:\Fafo\Data\Ethiopia_Uganda_2021\ETH\Sample". file handle DataPath /name "D:\Fafo\Data\Ethiopia_Uganda_2021\ETH\Survey1\Raw\AddisAbaba+Somali_PPS". * 4.1 Load data. * 4.1.1 Sample data. GET DATA /TYPE=XLSX /FILE='SampPath\Sample information for weigh calculations-Addis Ababa+Somila_PPS.xlsx' /SHEET=name 'Initial clusters' /CELLRANGE=FULL /READNAMES=ON /LEADINGSPACES IGNORE=YES /TRAILINGSPACES IGNORE=YES /DATATYPEMIN PERCENTAGE=95.0 /HIDDEN IGNORE=YES. dataset name Samp. delete variables GEOCode1 GEOName1 GEOCode2 GEOName2 GEOCode3 GEOName3 GEOCode4 GEOName4 GEOCode5 GEOName5 GEOCode6 GEOName6 GEOName6 Otherinformationofcluster Urbanorrural1urban2rural Campornoncamp1camp2noncamp Country1Ethiopia2Uganda NumberofEAsinStratum. rename variables ( ClusterID = QHCLUSTER ) ( ActualClusterID = InitCluster ) ( NumberofselectedEAs = ClusSelStra ) ( Numberofhhsinstratum = HHStra ) ( NumberofhhsBEFOREListing = HHClusBe ) ( NumberofnationalhhsAFTERlisting = HHClusList_Nat ) ( Samplesizeofselectednationalhhs = HHSelClus_Nat ) ( Numberofrefugeesfromlisting = HHClusList_Ref ) ( Samplesizeofselectedrefugeehhs = HHSelClus_Ref ). * remove duplicate cluster information. dataset declare Samp2. dataset activate Samp. aggregate outfile = Samp2 /break = InitCluster /Stratum = first( Stratum ) /ClusSelStra = first( ClusSelStra ) /HHStra = first( HHStra ) /HHClusBe = first( HHClusBe ) /HHClusList_Nat = first( HHClusList_Nat ) /HHClusList_Ref = first( HHClusList_Ref ) /HHSelClus_Nat = first( HHSelClus_Nat ) /HHSelClus_Ref = first( HHSelClus_Ref ). dataset close Samp. sort cases by InitCluster. * 4.1.2 Survey data. get file "DataPath\H_MAIN.sav" /keep QHCLUSTER QHID QHRESULT QHMEMBER QHRSIELIG QHRSI QHTYPE QHSTOP LS515 QHGEOCODE1. dataset name Main. sort cases by QHCLUSTER QHID. get file "DataPath\I_MAIN.sav" /keep QICLUSTER QIHID QIIDV QIRESULT QITYPE QISTOP SF1111 QIRESULTO /rename( QICLUSTER = QHCLUSTER ) ( QIHID = QHID ). dataset name IDV. sort cases by QHCLUSTER QHID QIIDV. match files file = main /file = IDV /by QHCLUSTER QHID. exe. dataset name Main2. dataset close Main. dataset close IDV. dataset activate main2. * Delete duplicate cases. if QHCLUSTER = 1087 and ( range( QHID, 4067, 4070 ) or range( QHID, 4072, 4085 ) ) delcase = 1. if QHCLUSTER = 1088 and any( QHID, 4086, 4087, 4088, 4089, 4090, 4092, 4094, 4095, 4096, 4097, 4098 ) delcase = 1. if QHCLUSTER = 1114 and range( QHID, 4182, 4213 ) delcase = 1. if QHCLUSTER = 1128 and range( QHID, 4266, 4306 ) delcase = 1. if QHCLUSTER = 1142 and ( range( QHID, 4393, 4397 ) or any( QHID, 4400, 4403, 4404, 4408, 4409, 4410, 4411, 4415, 4416, 4417 ) ) delcase = 1. if QHCLUSTER = 1144 and range( QHID, 4433, 4441 ) delcase = 1. if QHCLUSTER = 911263 and range( QHID, 5507, 5523 ) delcase = 1. select if missing( delcase ). * Change cluster ID. compute InitCluster = QHCLUSTER. If QHCLUSTER = 108711 InitCluster = 1087 . If QHCLUSTER = 108811 InitCluster = 1088 . If QHCLUSTER = 111411 InitCluster = 1114 . If QHCLUSTER = 112811 InitCluster = 1128 . If QHCLUSTER = 114211 InitCluster = 1142 . If QHCLUSTER = 114411 InitCluster = 1144 . If QHCLUSTER = 910871 InitCluster = 10871 . If QHCLUSTER = 910872 InitCluster = 10872 . If QHCLUSTER = 910873 InitCluster = 10873 . If QHCLUSTER = 910874 InitCluster = 10874 . If QHCLUSTER = 910875 InitCluster = 10875 . If QHCLUSTER = 910881 InitCluster = 10881 . If QHCLUSTER = 910882 InitCluster = 10882 . If QHCLUSTER = 910883 InitCluster = 10883 . If QHCLUSTER = 910884 InitCluster = 10884 . If QHCLUSTER = 910885 InitCluster = 10885 . If QHCLUSTER = 910886 InitCluster = 10886 . If QHCLUSTER = 910887 InitCluster = 10887 . If QHCLUSTER = 910991 InitCluster = 10991 . If QHCLUSTER = 910992 InitCluster = 10992 . If QHCLUSTER = 910993 InitCluster = 10993 . If QHCLUSTER = 910994 InitCluster = 10994 . If QHCLUSTER = 910995 InitCluster = 10995 . If QHCLUSTER = 910996 InitCluster = 10996 . If QHCLUSTER = 910997 InitCluster = 10997 . If QHCLUSTER = 911101 InitCluster = 11101 . If QHCLUSTER = 911102 InitCluster = 11102 . If QHCLUSTER = 911103 InitCluster = 11103 . If QHCLUSTER = 911104 InitCluster = 11104 . If QHCLUSTER = 911105 InitCluster = 11105 . If QHCLUSTER = 911106 InitCluster = 11106 . If QHCLUSTER = 911107 InitCluster = 11107 . If QHCLUSTER = 911131 InitCluster = 11131 . If QHCLUSTER = 911132 InitCluster = 11132 . If QHCLUSTER = 911133 InitCluster = 11133 . If QHCLUSTER = 911134 InitCluster = 11134 . If QHCLUSTER = 911135 InitCluster = 11135 . If QHCLUSTER = 911136 InitCluster = 11136 . If QHCLUSTER = 911137 InitCluster = 11137 . If QHCLUSTER = 911141 InitCluster = 11141 . If QHCLUSTER = 911142 InitCluster = 11142 . If QHCLUSTER = 911143 InitCluster = 11143 . If QHCLUSTER = 911144 InitCluster = 11144 . If QHCLUSTER = 911145 InitCluster = 11145 . If QHCLUSTER = 911146 InitCluster = 11146 . If QHCLUSTER = 911261 InitCluster = 11261 . If QHCLUSTER = 911262 InitCluster = 11262 . If QHCLUSTER = 911263 InitCluster = 11263 . If QHCLUSTER = 911264 InitCluster = 11264 . If QHCLUSTER = 911282 InitCluster = 11282 . If QHCLUSTER = 911283 InitCluster = 11283 . If QHCLUSTER = 911284 InitCluster = 11284 . If QHCLUSTER = 911285 InitCluster = 11285 . If QHCLUSTER = 911291 InitCluster = 11291 . If QHCLUSTER = 911292 InitCluster = 11292 . If QHCLUSTER = 911293 InitCluster = 11293 . If QHCLUSTER = 911294 InitCluster = 11294 . If QHCLUSTER = 911295 InitCluster = 11295 . If QHCLUSTER = 911296 InitCluster = 11296 . If QHCLUSTER = 911297 InitCluster = 11297 . If QHCLUSTER = 911353 InitCluster = 11353 . If QHCLUSTER = 911354 InitCluster = 11354 . If QHCLUSTER = 911355 InitCluster = 11355 . If QHCLUSTER = 911357 InitCluster = 11357 . If QHCLUSTER = 911358 InitCluster = 11358 . If QHCLUSTER = 911361 InitCluster = 11361 . If QHCLUSTER = 911362 InitCluster = 11362 . If QHCLUSTER = 911363 InitCluster = 11363 . If QHCLUSTER = 911364 InitCluster = 11364 . If QHCLUSTER = 911422 InitCluster = 11422 . If QHCLUSTER = 911423 InitCluster = 11423 . If QHCLUSTER = 911424 InitCluster = 11424 . If QHCLUSTER = 911425 InitCluster = 11425 . If QHCLUSTER = 911431 InitCluster = 11431 . If QHCLUSTER = 911432 InitCluster = 11432 . If QHCLUSTER = 911433 InitCluster = 11433 . If QHCLUSTER = 911434 InitCluster = 11434 . If QHCLUSTER = 911435 InitCluster = 11435 . If QHCLUSTER = 9112631 InitCluster = 11263 . sort cases by InitCluster. * 4.2 Check sample file with survey data. dataset activate Main2. * Change HH type due to note from Tewodros. if QHCLUSTER = 5005 and QHID = 2001 TypeChange = 1. if QHCLUSTER = 5018 and QHID = 2002 TypeChange = 1. if QHCLUSTER = 5018 and QHID = 2003 TypeChange = 1. if QHCLUSTER = 5025 and QHID = 2004 TypeChange = 1. if QHCLUSTER = 5025 and QHID = 2005 TypeChange = 1. if QHCLUSTER = 5029 and QHID = 2006 TypeChange = 1. if QHCLUSTER = 5036 and QHID = 2007 TypeChange = 1. if QHCLUSTER = 5036 and QHID = 2008 TypeChange = 1. if QHCLUSTER = 5043 and QHID = 2009 TypeChange = 1. if QHCLUSTER = 5043 and QHID = 2010 TypeChange = 1. if QHCLUSTER = 5050 and QHID = 4001 TypeChange = 1. if QHCLUSTER = 5051 and QHID = 4002 TypeChange = 1. if QHCLUSTER = 5051 and QHID = 4003 TypeChange = 1. if QHCLUSTER = 5051 and QHID = 4004 TypeChange = 1. if QHCLUSTER = 5051 and QHID = 4005 TypeChange = 1. if QHCLUSTER = 5066 and QHID = 4006 TypeChange = 1. if QHCLUSTER = 5066 and QHID = 4007 TypeChange = 1. if QHCLUSTER = 5067 and QHID = 4008 TypeChange = 1. if QHCLUSTER = 5067 and QHID = 4009 TypeChange = 1. if QHCLUSTER = 5067 and QHID = 4010 TypeChange = 1. if QHCLUSTER = 5067 and QHID = 4011 TypeChange = 1. if QHCLUSTER = 5067 and QHID = 4012 TypeChange = 1. if QHCLUSTER = 5067 and QHID = 4013 TypeChange = 1. if QHCLUSTER = 5069 and QHID = 4014 TypeChange = 1. if QHCLUSTER = 5069 and QHID = 4015 TypeChange = 1. if QHCLUSTER = 5069 and QHID = 4016 TypeChange = 1. if QHCLUSTER = 5069 and QHID = 4017 TypeChange = 1. if QHCLUSTER = 5069 and QHID = 4018 TypeChange = 1. if QHCLUSTER = 5073 and QHID = 4064 TypeChange = 1. if QHCLUSTER = 5073 and QHID = 4065 TypeChange = 1. if QHCLUSTER = 5073 and QHID = 4066 TypeChange = 1. if QHCLUSTER = 5073 and QHID = 4067 TypeChange = 1. if TypeChange = 1 QHTYPE = 2. compute HH_national = QHTYPE = 1. compute HH_refugee = QHTYPE = 2. dataset declare check. aggregate outfile = check /break = InitCluster /Tot_N = sum( HH_national ) /Tot_R = sum( HH_refugee ). match files file =Samp2 /file = check /by InitCluster. exe. dataset name Samp3. dataset close Samp2. dataset close check. dataset activate Samp3. select if not missing( Stratum ). * 4.3 Inclusion Probability. * 4.3.1 PSU. dataset activate Samp3. compute Ppsu = HHClusBe * ClusSelStra / HHStra . * 4.3.2 Household. if HHClusList_Nat > 0 Phc_Nat = HHSelClus_Nat / HHClusList_Nat. if HHClusList_Ref > 0 Phc_Ref = HHSelClus_Ref / HHClusList_Ref. if HHSelClus_Nat = 0 and not missing( Phc_Nat ) Phc_Nat = $sysmis. if HHSelClus_Ref = 0 and not missing( Phc_Ref ) Phc_Ref = $sysmis. * For clusters that households were not selected separately. if any( InitCluster, 5076, 5077, 5078, 5079, 5080, 5081, 5082, 5083, 5084 ) Phc_Nat = ( HHSelClus_Nat + HHSelClus_Ref ) / ( HHClusList_Nat + HHClusList_Ref ). if any( InitCluster, 5076, 5077, 5078, 5079, 5080, 5081, 5082, 5083, 5084 ) Phc_Ref = ( HHSelClus_Nat + HHSelClus_Ref ) / ( HHClusList_Nat + HHClusList_Ref ). compute Ph_Nat = Ppsu * Phc_Nat. compute Ph_Ref = Ppsu * Phc_Ref. * 4.3.3 RSI. dataset activate Main2. if QHRSIELIG > 0 Prsi = 1 / QHRSIELIG. * 4.4 Weights. * 4.4.1 Cluster non-response corrector. dataset activate Samp3. compute ClusNo = 0. if missing( HHClusList_Nat ) and missing( HHClusList_Ref ) ClusNo = 1. dataset declare ClusNo. aggregate outfile = ClusNo /break = Stratum ClusNo /HHClusBe = sum( HHClusBe ). dataset activate ClusNo. sort cases by Stratum ClusNo. CASESTOVARS /ID = Stratum /INDEX = ClusNo /GROUPBY = VARIABLE. if missing( HHClusBe.1.00 ) HHClusBe.1.00 = 0. compute ClusNRcorrector = 1 / ( HHClusBe..00 / ( HHClusBe..00 + HHClusBe.1.00 ) ). exe. delete variables HHClusBe..00 HHClusBe.1.00. dataset activate Samp3. sort cases by Stratum InitCluster. match files table = ClusNo /file = Samp3 /by Stratum. dataset name Samp4. exe. dataset close Samp3. dataset close ClusNo. dataset activate Samp4. sort cases by InitCluster. match files table = Samp4 /file = Main2 /by InitCluster. exe. dataset name meg. dataset close Samp4. dataset close Main2. * 4.4.2 Modify interview result. * HH. if any( QHRESULT, 3, 11 ) and not missing( ls515 ) QHRESULT = 1. if QHRESULT = 11 and QHSTOP <> "" QHRESULT = 3. if QHRESULT = 11 and QHSTOP = "" QHRESULT = 9. * RSI. if QIRESULT = 11 and not missing( SF1111 ) QIRESULT = 1. * 4.4.3 Adjustment of non-response. * HH. compute Hcomp = 0. compute HnoResp = 0. compute HnoExist = 0. compute HnoDeter = 0. if any( QHRESULT, 1,2 ) Hcomp = 1. if any( QHRESULT, 3,6,8 ) HnoResp = 1. if any( QHRESULT, 5,7 ) HnoExist = 1. if any( QHRESULT, 9 ) HnoDeter = 1. * RSI. compute Rcomp = 0. compute RnoResp = 0. compute RnoExist = 0. compute RnoDeter = 0. if any( QIRESULT, 1,2 ) Rcomp = 1. if any( QIRESULT, 3, 4, 5, 10,11 ) RnoResp = 1. if any( QIRESULT, 6 ) and QHRSIELIG > 1 RnoResp = 1. if any( QIRESULT, 6 ) and QHRSIELIG = 1 RnoExist = 1. aggregate outfile = * mode = addvariables /break Stratum QHTYPE /HHcomp = sum( Hcomp ) /HHnoResp = sum( HnoResp ) /HHnoExist = sum( HnoExist ) /HHnoDeter = sum( HnoDeter ). aggregate outfile = * mode = addvariables /break = Stratum /RSIcomp = sum( Rcomp ) /RSInoResp = sum( RnoResp ) /RSInoExist = sum( RnoExist ) /RSInoDeter = sum( RnoDeter ). compute HHnoDeterCorr = HHnoResp / ( HHnoResp + HHnoExist ). compute RSInoDeterCorr = RSInoResp / ( RSInoResp + RSInoExist ). * Non-response correction ratio. compute HHNoRespCorr = 1 / ( HHcomp / ( HHcomp + ( HHnoResp + HHnoDeter * HHnoDeterCorr ) ) ). compute RSINoRespCorr = 1 / ( RSIcomp / ( RSIcomp + ( RSInoResp + RSInoDeter* RSInoDeterCorr ) ) ). * 4.4.4 Weights. * 4.4.4.1 Expansion weights. if QHTYPE = 1 P_h = Ph_Nat. if QHTYPE = 2 P_h = Ph_Ref. compute HHexpweigh_S = 1 / P_h. compute HHexpweigh_A = HHexpweigh_S * HHNoRespCorr * ClusNRcorrector. compute RSIexpweigh_S = HHexpweigh_A * ( 1 / Prsi ). compute RSIexpweigh_A = RSIexpweigh_S * RSINoRespCorr. * Trim expansion weights. compute HHexpweigh2 = HHexpweigh_A. if range( QHRESULT, 3,9 ) HHexpweigh2 = $sysmis. aggregate outfile = * mode=addvariables /break = QHGEOCODE1 /SDHHexpweigh = sd( HHexpweigh2 ) /MEDHHexpweigh = median( HHexpweigh2 ). compute HHexpweigh = HHexpweigh2. if HHexpweigh2 > MEDHHexpweigh + 2 * SDHHexpweigh HHexpweigh = MEDHHexpweigh + 2 * SDHHexpweigh. dataset activate meg. compute RSIexpweigh2 = RSIexpweigh_A. if range( QIRESULT, 3,11 ) or RSIexpweigh_A = 0 RSIexpweigh2 = $sysmis. aggregate outfile = * mode=addvariables /break = QHGEOCODE1 /SDRSIexpweigh = sd( RSIexpweigh2 ) /MEDRSIexpweigh = median( RSIexpweigh2 ). compute RSIexpweigh = RSIexpweigh2. if RSIexpweigh2 > MEDRSIexpweigh + 2 * SDRSIexpweigh RSIexpweigh = MEDRSIexpweigh + 2 * SDRSIexpweigh. * 4.4.4.2 Relative weights. aggregate outfile = * mode = addvariables /break = QHGEOCODE1 /MeanHHexpweigh = mean( HHexpweigh ). aggregate outfile=* mode = addvariables /break = QHGEOCODE1 /MeanRSIexpweigh = mean( RSIexpweigh ). exe. compute HHrelweigh = HHexpweigh / MeanHHexpweigh. compute RSIrelweigh = RSIexpweigh / MeanRSIexpweigh. exe. * 4.5 Save weights. compute PSU = InitCluster. variable labels QHCLUSTER "Cluster ID in survey field work" /Stratum "Stratum" /PSU "Primary sampling unit for PPS cluster sampling" /HHexpweigh "Household expansion weights for PPS cluster sampling" /HHrelweigh "Household relative weights for PPS cluster sampling" /RSIexpweigh "RSI expasion weights for PPS cluster sampling" /RSIrelweigh "RSI relative weights for PPS cluster sampling". save outfile "SampPath\Weight_ETH_PPS.sav" /keep QHCLUSTER QHID QIIDV Stratum PSU HHrelweigh RSIrelweigh HHexpweigh RSIexpweigh /rename ( PSU = PSU_PPS ) ( HHrelweigh = HHrelweigh_PPS ) ( RSIrelweigh = RSIrelweigh_PPS ) ( HHexpweigh = HHexpweigh_PPS ) ( RSIexpweigh = RSIexpweigh_PPS ). dataset close all.