FINAL QUALITY REPORT EU-SILC-2006 Slovenia

Transcription

1 REPUBLIC OF SLOVENIA FINAL QUALITY REPORT EU-SILC-2006 Slovena Report prepared by: Rhard Inglč Rud Seljak Matja Remec Martna Stare Stanka Inthar Document created: 24/11/2008, Last updated: 21/01/2009 1/56

2 CONTENTS 1 COMMON LONGITUDINAL EU INDICATORS COMMON LONGITUDINAL EUROPEAN UNION INDICATORS BASED ON THE LONGITUDINAL COMPONENT OF EU-SILC ACCURACY SAMPLE DESIGN Type of samplng desgn (stratfed, mult-stage, clustered) Samplng unts (one stage, two stages) Stratfcaton and substratfcaton crtera Sample sze and allocaton crtera... 4 In 2005 we obtaned 8314 responses, whch we keep n the survey n The total sample sze n 2006 s thus 8314 unts from prevous year plus 4203 unts n the new wave,.e persons Sample selecton schemes Sample dstrbuton over tme Renewal of sample: rotatonal groups Weghtng Substtutons SAMPLING ERRORS Standard error and effectve sample sze NON-SAMPLING ERRORS Samplng frame and coverage errors Measurement and processng errors Non-response errors MODE OF DATA COLLECTION IMPUTATION PROCEDURE IMPUTED RENT COMPANY CARS COMPARABILITY BASIC CONCEPTS AND DEFINITIONS COMPONENTS OF INCOME Dfferences between the natonal defntons and standard EU-SILC defntons, and an assessment of the consequences of the dfferences mentoned wll be reported for the followng target varables The source of procedure used for the collecton of ncome varable The form n whch ncome varables at component level have been obtaned The method used for obtanng ncome target varables n the requred form TRACING RULES COHERENCE THE DIFFERENCES BETWEEN HBS AND EU-SILC THE DIFFERENCES BETWEEN LFS AND EU-SILC THE DIFFERENCES BETWEEN EU-SILC 2005 AND /56

3 1 Common longtudnal EU ndcators 1.1 Common longtudnal European Unon ndcators based on the longtudnal component of EU-SILC EU-SILC was conducted the frst tme n 2005 and because of ths we could not calculate longtudnal ndcators. 2 Accuracy 2.1 Sample desgn Type of samplng desgn (stratfed, mult-stage, clustered) The sample desgn for Slovenan EU-SILC 2006 was two-stage stratfed desgn. In each stratum prmary samplng unts (PSUs) were frstly systematcally selected, and n the second stage 7 persons were selected n each PSU. We have used rotatonal desgn, meanng that three waves were preserved from the prevous year and just one wave was addtonally selected usng the descrbed desgn Samplng unts (one stage, two stages) In the frst stage samplng unts were selected, whch are clusters of enumeraton areas, whch are approxmately of the same sze, and then n the second stage 7 persons were selected n the selected PSUs. Unt of observaton are selected persons lvng n prvate households n Slovena and ther households. The data are collected from all household members who were on 31 st December 2005 aged 16 years or more. The selected person s also the sample person; other household members are not sample persons Stratfcaton and substratfcaton crtera The samplng frame of persons aged 16 years or more s dvded nto 6 strata, whch are defned accordng to the sze of the settlement and the proporton of agrcultural households n the settlement: 1. The frst stratum ncludes settlements wth fewer then nhabtants and wth less then 30% of agrcultural households; 2. The second stratum ncludes settlements wth fewer then nhabtants and wth at least 30% agrcultural households; 3. The thrd stratum ncludes settlements whch have from to nhabtants; 4. The fourth stratum ncludes settlements whch have from to nhabtants; 5. The ffth stratum s Marbor (the second largest cty n Slovena wth approx nhabtants); 6. The sxth stratum s Ljubljana (Slovena s captal wth approx nhabtants). 3/56

4 When selectng the samplng unts, explct stratfcaton accordng to the type of settlement was used (6 strata). Snce we wanted to mantan regonal representatveness, mplct stratfcaton accordng to statstcal regon was appled. It means that the lst of unts wthn strata was sorted accordng to statstcal regons. In Slovena there are 12 statstcal (NUTS3) regons: 1. Pomurska 2. Podravska 3. Koroška 4. Savnjska 5. Zasavska 6. Spodnjeposavska 7. Jugovzhodna Slovenja 8. Osrednjeslovenska 9. Gorenjska 10. Notranjsko-kraška 11. Gorška 12. Obalno-kraška Sample sze and allocaton crtera In Eurostat s document SILC/138/04 Framework Regulaton; Annex 2 on Sample Szes, the mnmal net sample sze s defned accordng to dfferent sample desgn schemes. Snce n Slovena we have a sample of persons, but n the household only the selected person s the sample person who responds to Socal varables, we have to obtan responses from at least 6750 selected persons and ther households. Snce n 2005 the nonresponse was larger then expected, the new part of the sample was enlarged for 700 unts. The samplng frame was dvded nto 6 strata. In 2005 we obtaned dfferent ntervewng rates n the strata; therefore we decded to oversample strata where we expected lower response rates. For oversamplng the data from the EU-SILC n 2005 were used. Table 1 shows how the structure alters because of the oversamplng of some strata. Table 1: Dstrbuton of the settlements n sx strata accordng to the number of nhabtants and the proporton of rural households n the settlement Strata, dstrbuton of settlements Structure Intervewng rate n EU-SILC 2005 Altered structure due to oversamplng Fewer then 2000 nhab., not rural 29.4% % Fewer than 2000 nhab., rural 23.5% % From 2000 to nhab. 16.1% % From to nhab. 13.3% % Marbor 4.8% % Ljubljana 12.9% % 4/56

5 On the frst stage 600 samplng unts were selected, and then n each samplng unt 6 to 8 persons aged 16 years or more were selected. The selected persons defne the households whch we wanted to ntervew. The sample sze of the new part of the sample was thus 4203 persons. In 2005 we obtaned 8314 responses, whch we keep n the survey n The total sample sze n 2006 s thus 8314 unts from prevous year plus 4203 unts n the new wave,.e persons Sample selecton schemes The samplng frame was dvded nto 6 strata and each stratum was sorted by 12 statstcal regons. Ths way we mplctly stratfed the sample also by statstcal regon. Wthn each stratum we systematcally selected 600 samplng unts, and then n each samplng unt 7 persons were selected. Persons aged 16 years were oversampled. In each samplng unt, persons aged 16 years and others were separately selected. a number of prmary samplng unts (= 600) b number of persons, who are selected n PSU (= 7) p proporton of persons aged 16 n PSU b 1 number of persons aged 16 who are selected n PSU b 2 number of persons aged 17 or more who are selected n PSU Probablty of selecton of person aged 16 n PSU I s Probablty of selecton of person aged 17 or more n PSU s Condtons: an. b1 an b 2 N pn b = b 1 + b 2 = ( ) N (1 p) N s proporton of persons aged 16 n the populaton,. an b1 N an pn. b 2 N (1 p) N We obtan a unquely solvable system of two lnear equatons wth two unknowns. Thus n the selected samplng unt we select: b b pb = 16-years olds and (1 + p) ( p) b (1 + p) = persons, aged 17 or more. Beacause of decmal number of selected persons n PSU (b 1, b 2 ), sze of PSUs s between 6 and 8. Therefore the fnal sample sze s 4203 persons. Probablty of selecton of person aged 16 n the PSU s: 5/56

6 an. N pb (1 + p) pn = an. N b (1 + p) N Probablty of selecton of person aged 17 or more n the PSU s: an. N ( p) b (1 + p)(1 p) N = an. N ( p) b (1 p ) N Sample dstrbuton over tme Feldwork for CAPI ntervewng lasted from 1 st February untl 5 th June 2006 and for CATI ntervewng lasted from 1 st February untl 30 th March. In 2005 ntervewng lasted from February untl June Renewal of sample: rotatonal groups The samplng frame has a four-year rotatonal desgn. Persons and ther households reman n the sample for four years or four waves; each year one quarter of the sample s replaced. One quarter of the sample s dropped and one quarter s added each year. Each quarter of the sample s called a rotatonal group and has to be representatve for the target populaton. Snce we had n 2005 a lower ntervewng rate than expected, we had to enlarge the sample for 2006; otherwse our sample sze would be too small after four years for longtudnal analyss. In 2006 we should have dropped out the fourth wave from 2005, but we have decded to keep the fourth wave and dvde t nto three parts and reallocate them to the remanng three waves form Therefore all households whch responded n 2005 were n 2006 ntervewed agan. Snce we have decded ths before data processng of the 2005 survey, we have renumbered ntally selected samplng unts n the way that we have nstead of four three rotatonal groups. None of the rotatonal groups were dropped out n In 2006 only one new rotatonal group was added, so that we have four rotatonal groups n Table 2: Number of PSU and number of selected persons Sample n 2005 dvded nto 3 rotatonal groups: Rotatonal group Number of PSUs Number of selected persons Total /56

7 Sample n 2006 dvded nto 4 rotatonal groups Rotatonal group Number of PSUs Number of selected persons Total Wth renumberng of the rotatonal groups we enlarged the sze of each group, and therefore also the new part of the sample n 2006 s larger. The followng years we wll mantan 4 rotatonal groups: one group wll be dropped off and one wll be added. Therefore we wll have 75% of overlap of the sample Weghtng The crossectonal weghts for the frst wave were calculated dfferently as those for the consecutve waves Cross-sectonal weghts for the frst wave The weghts were calculated n three consecutve steps. In the frst step the samplng weght (desgn factor), n the second the non-response adjustment factor and n the thrd the calbraton factor was calculated. The fnal weght was the product of all three factors. The weghts were calculated for the selected household (selected person of the household) and for all the persons ncluded n the survey. In EU-SILC the sample of persons aged 16 years or more was selected from the Central Regster of Populaton. Sample persons and ther households were ntervewed Desgn factor The samplng weght for the sample person PB070 s nversely proportonal to the probablty of selecton and the weght s calculated when the person s selected n the sample. For the persons that were n the sample also n the prevous year, the samplng weght s taken from the prevous year, yet the samplng weghts are to be calculated just for the persons that are new n the sample. Snce the PPS 2-stage samplng was used, the samplng weght for the N h selected person n the partcular stratum ( h ), can smple be calculated as w h =, where N h n s the stratum numbers of the persons n the samplng frame and n h s the stratum numbers of the persons n the sample. The samplng weght of the household of the selected person: DB080 Snce SORS doesn t yet have a regster of households, the selecton of the household s done wth the selecton of the person. Snce households wth more persons aged 16 years or more have a larger probablty of selecton then smaller households, ths has to be corrected wth weghtng n such a way that all households have equal probablty of beng selected n the sample. Thus the probablty of selecton of the household s equal to the probablty of h 7/56

8 selecton of the person dvded by the number of elgble persons (aged 16+) n the household M: DB080=PB070 / M h The samplng weght for the households has to be calculated for all households n the sample, not only for the respondng households. Snce for the households that dd not respond we do not know ther sze, we have calculated the average sze of the household of persons aged 16 or more accordng to dfferent statstcal regons and type of settlement (47 classes) and we mputed ths value to households that dd not respond. Thus we could calculate the probablty of selecton also for households that dd not respond Non-response adjustments The non-response factor was calculated for each stratum. Frst the sample was dvded nto three categores: responses, non-responses and out-of-scope unts. The non-response adjustment factor s calculated: the stratum and w NR r nr nh + nh =, where n n number of the non- responses n the stratum. nr h r h r n h s the number of the responses n Adjustments to external data (level, varables used and sources) The fnal step of the calculaton of the weghts was the calculaton of the calbraton factors. By the calbraton procedures the weghted sums of some key varables are set to the known populaton values. These populaton values are obtaned from the dfferent admnstratve sources. For the calbraton of weghts we used SAS Macro Calmar. We performed calbraton for the level of households, as well as for the level of the persons. For the calbraton we used: 1. for households: - Famly and chldren related allowance (HY050) from the admnstratve source for famly and chldren related allowances 2. for persons: - Sex- age classes dstrbuton from the Central Regster of Populaton - Employee cash or near cash ncome mnus sckness benefts from the admnstratve source for ncomes - Pensons from the admnstratve sources for pensons - Unemployment benefts (PY090) from the admnstratve source for unemployment benefts - Educaton related allowances from the statstcal source about scholarshps 8/56

9 Fnal cross-sectonal weghts The cross-sectonal weght for the household (DB090) s equal to the calbrated weght. The sum of weghts s equal to the sum of the estmated number of households n Slovena. Wth the selected person also the household whch has to be ntervewed s defned. All household members have the same weght, ths s the cross-sectonal weght. The crosssectonal weght of the person RB050, whch all persons get n the household regster, and the cross-sectonal weght of persons aged 16 years or more PB040 n the person regster are equal to the cross-sectonal weght of the household. RB050= PB040=DB090 The cross-sectonal weght for the selected person PB060 s equal to the cross-sectonal weght of the household of ths person multpled by the number of persons aged 16+: PB060= DB090 * M h The cross-sectonal weght for chldren who were younger than 13 years on 31 st December 2005 s RL070. Weghts are calculated n ths way that we calculate for each age group a factor: f =number of chldren n the populaton/weghted number of chldren n the survey, =1,2,,12. Wth ths factor we multply the cross-sectonal weght RB050 of a chld n the correspondng age group. RL070=f *RB050, =1,2,,12 The base weghts for the persons n the frst wave are equal to the cross-sectonal weghts for the persons Cross-sectonal weghts for the consecutve waves Base weghts The Base weghts for the persons were calculated by takng the base weghts from the prevous year and then adjust these weghts for the attrton n the Sex- age classes. Usng the weght-share method we then calculated the weghts for the mmgrants, re-entres and newborns. After that for each of the rotatonal groups the weghts were adjusted to the adequate longtudnal populaton counts n each Sex- age class Fnal cross-sectonal weghts The cross-sectonal weghts for the households were calculated by frstly takng the average of the base weghts for the belongng persons and then calbrate these weghts for each rotatonal group to the same margn values as used n The cross-sectonal weghts for the persons and selected persons were calculated by the same procedure as used for the frst wave. 9/56

10 Longtudnal weghts The longtudnal weghts were calculated by takng the base weghts and then calbrate these weghts to the Sex-age structure of the correspondng longtudnal populaton whch was determned as the overlap of the regster populaton n the consecutve years Substtutons In EU-SILC we dd not have substtute unts. 2.2 Samplng errors Standard error and effectve sample sze Table 3: The mean, the total number of observatons (before and after mputatons) and the standard errors cross sectonal EU-SILC 2005 Income components HY010 Descrpton Mean (weghted) Number of observatons before mputatons (n the survey wth value not equal 0 before mputatons) Number observatons after mputatons (n survey) of the Standard errors Total gross household ncome HY020 Total dsposable household ncome HY022 Total dsposable household ncome before socal transfers except old age and survvor's benefts HY023 Total dsposable household ncome before socal transfers ncludng old-age and survvor's benefts HY040G HY040N Income from rental of a property or land gross Income from rental of a property or land net HY090G Interest, dvdends, proft form captal nvestments n unncorporated busness HY090N Interest, dvdends, proft form captal /56

11 nvestments n unncorporated busness HY050G Famly/Chldren related allowances HY050N Famly/Chldren related allowances HY060G Socal excluson not elsewhere classfed HY060N Socal excluson not elsewhere classfed HY070G Housng allowances HY070N Housng allowances HY080G Regular nter household cash transfer receved gross HY080N Regular nter household cash transfer receved net HY100G Interest repayments on mortgage gross HY100N Interest repayments on mortgage net HY110G Income receved by people aged under 16 gross HY110N Income receved by people aged under 16 net HY120G Regular taxes on wealth gross HY120N Regular taxes on wealth net HY130G Regular nter household cash transfer pad gross HY130N Regular nter household cash transfer pad - net HY140G tax on ncome and socal contrbuton HY140N tax on ncome and socal contrbuton HY145N Repayments/recepts for tax adjustment Varable Descrpton Mean (weghted) Number of observatons before mputatons (n the survey wth value not equal 0) PY010G Employee cash or near cash ncome Number of observatons after mputatons (n the survey) Standard errors gross PY010N Employee cash or near cash ncome net PY020G Non-Cash employee /56

12 ncome net PY020N Non-Cash employee ncome net PY035G Contrbutons to ndvdual prvate pensons plans gross PY035N Contrbutons to ndvdual prvate pensons plans gross PY050G Cash benefts or losses PY050N from self-employment Cash benefts or losses from self-employment PY070G Value of goods produced by own consumpton PY070N Value of goods produced by own consumpton PY080G Penson from ndvdual prvate plans gross PY080N Penson from ndvdual prvate plans net PY090G Unemployment benefts gross PY090N Unemployment benefts net PY100G Old age benefts gross PY100N Old age benefts net PY110G Survvor benefts net PY110N Survvor' age benefts gross PY120G Sckness benefts gross PY120N Sckness benefts net PY130G Dsablty benefts gross PY130N Dsablty benefts net PY140G Educaton related allowances gross PY140N Educaton related allowances net /56

13 Table 4: The mean, the total number of observatons (before and after mputatons) and the standard errors cross sectonal EU-SILC 2006 Income components HY010 Descrpton Mean (weghted) Number of observatons before mputatons (n the survey wth value not equal 0 before mputatons) Number observatons after mputatons (n survey) of the Standard errors Total gross household ncome HY020 Total dsposable household ncome HY022 Total dsposable household ncome before socal transfers except old age and survvor's benefts HY023 Total dsposable household ncome before socal transfers ncludng old-age and survvor's benefts HY040G HY040N Income from rental of a property or land gross Income from rental of a property or land net HY090G Interest, dvdends, proft form captal nvestments n unncorporated busness HY090N Interest, dvdends, proft form captal HY050G HY050N HY060G HY060N nvestments n unncorporated busness Famly/Chldren related allowances Famly/Chldren related allowances Socal excluson not elsewhere classfed Socal excluson not elsewhere classfed HY070G Housng allowances HY070N Housng allowances HY080G Regular nter household cash transfer receved gross HY080N Regular nter household cash transfer receved net HY100G Interest repayments on mortgage gross /56

14 HY100N Interest repayments on mortgage net HY110G Income receved by people aged under 16 gross HY110N Income receved by people aged under 16 net HY120G Regular taxes on wealth gross HY120N Regular taxes on wealth net HY130G Regular nter household cash transfer pad gross HY130N Regular nter household cash transfer pad - net HY140G tax on ncome and socal contrbuton HY140N tax on ncome and socal contrbuton HY145N Repayments/recepts for tax adjustment Varable Descrpton Mean (weghted) Number of observatons before mputatons (n the survey wth value not equal 0) PY010G Employee cash or near cash ncome PY010N Number of observatons after mputatons (n the survey) Standard errors gross Employee cash or near cash ncome net PY020G Non-Cash employee ncome net PY020N Non-Cash employee ncome net PY035G Contrbutons to ndvdual prvate pensons plans gross PY035N Contrbutons to ndvdual prvate pensons plans gross PY050G Cash benefts or losses from self-employment PY050N Cash benefts or losses from self-employment PY070G Value of goods produced by own consumpton PY070N Value of goods produced by own consumpton PY080G Penson from /56

15 ndvdual prvate plans gross PY080N Penson from ndvdual prvate plans net PY090G Unemployment benefts gross PY090N Unemployment benefts net PY100G Old age benefts gross PY100N Old age benefts net PY110G Survvor benefts net PY110N Survvor' age benefts gross PY120G Sckness benefts gross PY120N Sckness benefts net PY130G Dsablty benefts gross PY130N Dsablty benefts net PY140G Educaton related allowances gross PY140N Educaton related allowances net /56

16 Table 5: The mean, the number of observatons (before and after mputatons) and the standard error for the equvalsed dsposable ncome breakdown by sex, age groups and household sze: Equvalsed dsposable ncome Mean Number of Standard error observatons after mputatons 1 household member household members household members and more household members <25 years Male Female Table 6: Standard errors and acheved sampled sze for some ndcators were calculated by usng the Bootstrap replcaton method, EU-SILC 2006 Indcator Value Acheved sample sze Confdence Interval at 95% Standard error Lower Upper CV(%) At-rsk-of-poverty rate after socal transfers total 11, ,29 11,1 12,3 2,46 At-rsk-of-poverty rate after socal transfers men total 10, ,31 9,7 10,9 3,01 At-rsk-of-poverty rate after socal transfers - women total 13, ,35 12,3 13,7 2,71 At-rsk-of-poverty rate after socal transfers -age group , ,67 10,5 13,1 5,71 At-rsk-of-poverty rate after socal transfers -age group , ,27 11,1 12,1 2,36 At-rsk-of-poverty rate after socal transfers -age group , ,31 9,5 10,7 3,02 At-rsk-of-poverty rate after socal transfers -age group , ,71 18,6 21,4 3,57 Inequalty of ncome dstrbuton S80/S20 ncome quntle share rato 3, ,09 3,2 3,6 2,61 Before socal transfers except old-age and survvors' benefts At-rsk-of-poverty rate before socal transfers - total 24, ,33 23,6 24,8 1,35 At-rsk-of-poverty rate before socal transfers - men total 22, ,37 22,2 23,6 1,63 At-rsk-of-poverty rate before socal transfers - women total 25, ,37 24,7 26,1 1,47 At-rsk-of-poverty rate before socal transfers -age group , ,80 24,7 27,9 3,03 At-rsk-of-poverty rate before socal transfers-age group , ,31 23,2 24,4 1,31 At-rsk-of-poverty rate before socal transfers -age group , ,79 30,5 33,7 2,47 Before socal ncludng old-age and survvors' benefts At-rsk-of-poverty rate before socal transfers - total 40, ,33 40,0 41,4 0,82 At-rsk-of-poverty rate before socal transfers - men total 38, ,36 37,5 38,9 0,95 At-rsk-of-poverty rate before socal transfers - women total 43, ,40 42,3 43,9 0,92 At-rsk-of-poverty rate before socal transfers -age group , ,77 28,7 31,7 2,56 At-rsk-of-poverty rate before socal transfers-age group , ,32 42,0 43,2 0,76 At-rsk-of-poverty rate before socal transfers -age group , ,75 82,1 85,1 0,89 Gn coeffcent 23, ,63 22,6 25,0 2,63 Mean equvalsed dsposable ncome ,04 16/56

17 2.3 Non-samplng errors Samplng frame and coverage errors The bass for the samplng frame s the Central Regster of Populaton (CRP), whch s lnked to the Regster of Terrtoral Unts. The samplng frame consttutes persons aged 16 years or more on 31 st of December Besdes the CRP we also use the frame of enumeraton areas. Snce some enumeraton areas do not have enough nhabtants, those enumeraton areas were lnked wth neghbourng areas nto larger terrtoral unts.e. samplng unts, whch were the samplng frame n the frst stage. The qualty of the CRP s dffcult to measure, snce the Census and the CRP are based on dfferent methodologes. Whle n the Census all persons lvng at the address at least one year are counted, current statstcs counts n the populaton persons who are regstered n Slovena and lve n Slovena at least three months. Therefore n the Census 2002 there are almost fewer persons than n the CRP (1.55%). The dscrepancy between the Census and the CRP s 1.72%. In the CRP are also persons who moved out of Slovena (temporarly or for good), but have not reported ths to the authortes. When desgnng the samplng frame we dd not have n the frame foregners who lve n Slovena and are by defnton the populaton of Slovena. There are approxmately foregners n Slovena. Therefore we have approxmately 2% of undercoverage n the samplng frame. Also we do not have the data n the CRP whch persons are lvng n collectve households. Accordng to the Census 2002 there are approxmately such persons. The CRP s daly updated, but SORS obtans the database every three months whch s a cross-secton of the CRP on a certan date. Therefore the CRP we work wth s 3 months old. For EU-SILC the samplng frame was bult from the CRP on 30 th June Before the feldwork we updated the samplng frame wth the latest avalable CRP data at the Mnstry of the Interor; so we have excluded form the feldwork persons who have ded or moved abroad as non-response. In case that a person has changed the address, the ntervewer was sent to the new address, but we mantaned varables that defne sample desgn at the old address. From the CRP we have randomly selected persons aged 16 or more. At the addresses of selected persons the selected person and hs or her household were ntervewed. If selected persons dd not lve at the address from the CRP where they are regstered, we dd not follow them but we consdered ths as non-response. Households where nobody s regstered at that address were thus excluded from the samplng frame Measurement and processng errors Measurement errors As n most surveys, the questonnare can be one sources of potentonal measurement errors. Unsatsfactory organzaton and desgn of the survey may results n output dfferent to the realty. For the case of EU-SILC the wordng and phrasng of the questons can lead to msunderstandngs, also dfferent orderng of the questons can result n dfferent answers. The data are a combnaton of ntervews and regster nformaton. The ntervews are carred out by CATI or CAPI. (CATI: 52% and CAPI: 48 %). The general mode of collecton was personal ntervew of a selected person. The household respondent was chosen by the ntervewer as the one who had the best knowledge of the household s affars. For part of questons for selected person the ntervewers were nstructed to prefer ntervewng the selected person whenever possble. Some basc nformaton for households that were n 17/56

18 prevous wave we just verfed and so we dscharged households. As n all surveys there s hghly possble that ntervewer can nfluence on respondent's answers. Durng the collectng data phase we dd regular checks on ther progress. On CATI ntervewng we montored all the tme ntervewers and n the same tme we warned them about mstakes. In our studo we have possblty to lsten the ntervew and n the same tme we can see on the computer what ntervewer enter nto the computer. The ntervewers do not know when they are nspected. CAPI ntervewers are oblged to send to the Offce every fortnght the data whch they collected. We checked frequency of some key answers and f we found out that somethng unexpected happened wth sngle ntervewer we asked hm for the reasons. Before the feld work began we organzed lessons durng 15 th January 2007 and 31 th January At 25 th January tll 31 th January we organsed nne lessons for both CAPI and CATI ntervewers. Each ntervewer was oblged to partcpate n one of those lessons, whch were 2 tmes 4 hours long. In the frst part of the lesson we nstructed ntervewers about theory, at the second part we organzed practcal ntervewng n the groups wth 3 to 4 ntervewers wth lap-tops. We prepared the questonnare and answers n advance, that we can see f the ntervewer understands meanng of the questons. In the case that ntervewer was changed (do now wsh to be ntervewer, do not work accordng to nstructons), the addtonal lessons was organsed. At the same tme we had approxmately 60 CAPI ntervewers (most of them were experenced, but some ntervewers are not), and approxmately 25 CATI ntervewers (most of them students, whose almost all had experence wth callng n households.) For CATI ntervewers specal lessons was organsed whch have the smlar content as for CAPI ntervewers. Specal tranng was organzed also for controllers and other techncal stuff. On all tranngs we explaned the purpose of ths survey, the methodology, questonnares and organzatonal part as well. CAPI ntervewers got on the lessons advanced letters and they sent them ther self to the sampled households some days before they ntended vst the household. For the CATI ntervewng all advanced letters were sent by Offce two days before began the ntervewng. To all letters are added small leaflet wth the some results from prevous year, where t s possble to get results, etc. In the constructon of the Slovenan questonnare we both adapted queston and desgn from our LFS questonnare for personal questons (especally questons related to labour market) and HBS questonnare for household and expendture questons. The core of questonnare was bult accordng to the recommendatons of Eurostat. In some cases the phrasng of questons have n some way dverge from Eurostat recommendatons because of Slovenan standards. Here are lsted dfferences when comparng our questonnare and Eurostat recommendatons. In 2007 we changed all ncome varables from Slovenan tolars (SIT) to EUR. In the questonnare t s possble that ntervewee answered n SIT or n EURO. We ntroduce for all 18/56

19 these varables new varable for currency and after the feld work was fnshed we recalculate all ncome varables nto EUR. Not ncome varables: HH010 We had more categores, but all categores are easly translated to Eurostat categores. HH020 We had more categores, but all categores are easly translated to Eurostat categores. HH030 The room s defned as space wth at least 6 square meters. HH070 Total housng costs are asked wth several questons costs for cold water, costs for sewage removal, costs for refuse removal, heatng, contrbuton to reserve fund, nsurance, and nterest for mortgage, rent, and regular mantenance. We summed up all varables from these questons to get HH070. HS070 HS110 n our survey we added some other durables (vdeo recorder, DVD player, dgtal camera etc.). PB130, PB140 we collected these data wth the questonnare, but f the data were dfferentated accordng to the central regster of populaton, we took the data from the regster. PB190, PB210 ths data we took from regster of populaton. PB220A, PB220B data were collected by questonnare. PE040 the data are from Statstcal regster of employment for actve persons, for others we collect the data va questonnare. PH040 and PH060 the questons were splted nto two questons: AC4 Was there any tme when selected person durng the last 12 months when he/she really needed to consult a medcal specalst (except dentst)? 1. Yes AC5 2. No queston about need of the dentst. AC5 Dd selected person get a help of a medcal specalst? 1. Yes 2. No. PL020 The qeuston s from 2006 onward ncluded nto the qeustonnare. PL025 The queston s from 2006 onward ncluded for all household members nto the questonnare. PL030 The qeuston s from 2006 onward ncluded for all household members nto the questonnare. 19/56

20 PL040 The queston s from 2006 onward ncluded for all household members nto the qeustonnare. PL050 for actve persons we got the data about occupaton from the statstcal regster of employment. For nactve (selected) persons we asked the queston about occupaton n the questonnare. After conductng the survey, we coded the occupaton nto sco-88(com) accordng the descrpton of the occupaton. Codng s done by professonal coders who also do the codng n the LFS. PL060 The queston s from 2006 onward ncluded for all household members nto the qeustonnare. PL070-PL085 It was constructed from the statstcal regster of employment and from the regsters from Health Insurance Company. The qeustonnare s a source for students. PL087 It was constructed from PL070-PL085 and from the questonnare. PL090 The source for ths varable s regster from Health Insurance Company. PL100. The queston s from 2006 ncluded for all household members nto the qeustonnare. PL210A-PL210L Constructed from statstcal regster of employment and Health Insurance Company. We have state on the last day of each month. The source for students were qeustonnare. The data for persons whch are not n any regster or any other source, are mputed accordng to the data from last year. The datafle from Tax authorty was edted n advance. Before we began the data processsng wth eu-slc we checked the data from tax datafle. We edted mpossble values (for example negatve values) and some very extreme values. Some mputatons were also made n advnace we dd logcal check and n the case of nconsstency we mputed values. These mputatons are not ncluded nto the mputaton factor n eu-slc database. All other ncome fles (socal alowances, pensons etc.) were not edted n advance for whole populaton, but only for eu-slc populaton Processng errors The questonnare was programmed n Blase. Data entry controls were bult nto the electronc questonnare, and there was less need for post data control. Control of data n the programme was done n varous ways. All numerc varables had absolute lmts for data entry. We had several syntax checks, one of them were sgnals (soft errors) whch gave a warnng to the ntervewers f the answer was ether unlkely because t was extreme or because t dd not correspond to answer gven to questons asked earler. These sgnals could be overrdden f the answer n queston was confrmed. And smlar hard errors, whch t was mpossble to overrde. We also had several logcal checks. Here are examples of syntax checks and one logcal check: 20/56

21 Soft syntax error: Varable (PL060): Number of hours usually worked per week n man job: f ntervewer entered less then 8 or more then 70 hours there was a sgnal: Really less then 8 or more then 70 hours per week n man job? The answer could be yes suppress or no correct the number of hours. Hard syntax error: Varable HB080/HB090: Person 1 and Person 2 responsble for the accommodaton: f ntervewer entered two tmes the same person there was a hard error: Person 1 responsble for the accommodaton and Person 2 responsble for the accommodaton can not be same. Logcal error: Varable PL030: Self-defned current economc status: f ntervewer entered the person aged 16 and more s a preschool chld there was an error: The person s 16 or more year old so can not be a preschool chld. After checkng the data from all sources separately, we compose so called ntegrated database wth all the data. In the case of logcal mstakes and nconsstency of the data, we edted the data to the most probably value Non-response errors Acheved sample sze Both for households and for the ndvduals we were nterested what the acheved sample sze was. Snce we have the sample of persons, and the data are obtaned both from the ntervew and from the regsters, the household s counted to be ntervewed only f household questonnare s completed and f also questonnare for the selected person s completed. From other household members data are obtaned from regsters. Acheved sample sze s calculated for 1. Number of selected respondents who are members of the households for whch the ntervew s accepted for the database (DB135 = 1), and who completed a personal ntervew (RB250 = 11 to 13); 2. Number of persons 16 years or older who are members of the households for whch the ntervew s accepted for the database (DB135 = 1), and who completed a personal ntervew (RB250 = 11 to 13); 21/56

22 Table 7. Acheved sample sze for total and rotatonal group breakdown Longtudnal database Year HB Rotatonal group No. of selected respondents (sample persons) from who nformaton s completed from ntervews and regsters DB135 = 1 & RB250=13 No. of persons 16+ who are members of the households for whch the ntervew s accepted for the database and from who nformaton s completed only from regsters DB135 = 1 & RB250=12 No. of persons 16+ who are members of the households for whch the ntervew s accepted for the database DB135 = 1 & RB250=12,13 Total Total Unt non-response For the total sample, the unt non-response wll be calculated by removng, from the numerator and the denomnator of the formulas descrbed below, those unts that accordng to the tracng rules are out of scope. Household non-response rates (NRh) wll be computed as follows: Where NRh=(1-(Ra * Rh)) * 100 Ra s the address contact rate. DB120 s the record of contact at the address. The Ra s calculated as follows: Ra = = Condton that have to be fulflled that the household s accepted to household regster are completed both household and personal questonnares. In our survey there are 9478 such households. Varable measures proporton of households that are acceptable for the database. Percentage s calculated form elgble households on contacted addresses. 22/56

23 Rh s the proporton of complete household ntervews accepted for the database. DB130 s the household questonnare result, and DB135 s the household ntervew acceptance result Rh = = = Therefore NRh=(1-(Ra * Rh)) * 100=(1-0,79473)*100=20,527% Indvdual non-response rates (NRp) wll be computed as follows: Where NRp=(1-(Rp)) * 100 Rp s the proporton of complete personal ntervews wthn the households accepted for the database RB245 s the respondent status, and RB250 s the data status. For those Members States where a sample of persons rather than a sample of households (addresses) was selected, the ndvdual non-response rates wll be calculated for the selected respondent (RB245=2), for all ndvduals aged 16 years or older (RB245=2+3) and for the nonselected respondent (RB245=3). [ RB250 = 13] = [ RB245 = 2] 9474 Rp = = [ RB250 = ] = [ RB245 = 2 + 3] Rp = = for the selected respondent for all ndvduals aged 16 years or older 23/56

24 [ RB250 = 12] = [ RB245 = 3] Rp = = for the nonselected respondent Thus NRp=(1-(Rp)) * 100=0 for the selected respondent (RB245=2), for all ndvduals aged 16 years or older (RB245=2+3) and for the nonselected respondent (RB245=3). Overall ndvdual non-response rates (*NRp) wll be computed as follows: *NRp=(1-(Ra * Rh * Rp)) * 100 = ( * *1)*100 = Longtudnal response rates Households: Wave response rate Percentage of households successfully ntervewed (DB135 = 1) whch were passed on to wave t (from wave t-1) or newly created or added durng wave t, excludng those out of scope (under the tracng rules) or non-exstent. W_RR= 6580/8287=0.79 Longtudnal follow up rate Percentage of households whch are passed on to wave t+1 for follow-up wthn the households receved nto wave t from wave t-1, excludng those out of scope (under the tracng rules) or non-exstent. LF_R=6747/8287=0.81 Follow up rato Number of households passed on from wave t to wave t+1 n comparson to the number of households receved for follow-up at wave t from wave t-1. F_RAT=6747/8287=0.81 Acheved sample sze rato Rato of the number of households accepted for the database (DB135 = 1) n wave t to the number of households accepted for the database (DB135 = 1) n wave t-1. ASS_RAT=6580/8287=0.79 Persons: Wave response rate Percentage of sample persons successfully ntervewed (RB250 = 11,12,13) among those passed on to wave t (from wave t-1) or newly created or added durng wave t, excludng those out of scope (under the tracng rules). 24/56

25 W_RR_SP= 6580/6580=1 Percentage of co-resdents selected n wave 1 successfully ntervewed (RB = 11,12,13) among those passed on to wave t (from wave t-1). W_RR_C= 12359/15913=0.78 Longtudnal follow up rate Percentage of sample persons successfully ntervewed (RB250 = 11,12,13) n wave t out of all of sample persons selected, excludng those who have ded or been found nelgble (out of scope), breakdown by causes of non-response. LF_R_SP= 6580/6580=1 Acheved sample sze rato Rato of the number of completed personal ntervews (RB250 = 11,12,13) n wave t to the number of completed personal ntervews n wave t-1. Ths rato wll be defned for sample persons and for all persons ncludng non-sample persons aged 16+ and for co-resdents aged 16+ selected n frst wave. ASS_RAT_P=18939/23862=0.79 Response rate for non sample persons RR_NSP=12359/13034= Dstrbuton of households by household status (DB110), by record of contact at address (DB120), by household qeustonnare result (DB130) and by household ntervew acceptance (DB135) Table 8. Dstrbuton of orgnal unts by record of contact at address'. Rotatonal group and total cross sectonal 2005 Total Rotatonal group 1 Rotatonal group 2 Rotatonal group 3 Number % Number % Number % Number % Total (DB120 = 11 to 23) Address contacted (DB120 = 11) Address non-contacted (DB120 = 21 to 23) Total address non-contacted (DB120 = 21 to 23) Address cannot be located (DB120= 21) Address unable to access (DB120 = 22) Address does not exsts or s non-resdental address or s unoccuped or not prncpal resdence (DB120 = 23) /56

26 Table 9. Dstrbuton of address contacted by household questonnare result and by household ntervew acceptance. Rotatonal group and total cross sectonal 2005 Total Rotatonal group 1 Rotatonal group 2 Rotatonal group 3 Number % Number % Number % Number % Total Household questonnare completed (DB130 = 11) Intervew not completed (DB130 = 21 to 24) Total ntervew not completed (DB130 = 21 to 24) Refusal to co-operate (DB130 = 21) Entrely household temporarly away for duraton of feldwork (DB130 = 22) Household unable to respond (llness, ncapacty, etc.) (DB130 = 23) Other reasons (DB130 = 24) Household questonnare completed (DB135=1+2) Intervew accepted for data base (DB135 = 1) Intervew rejected (DB135 = 2) /56

27 Table 10. Dstrbuton of orgnal unts by record of contact at address'. Rotatonal group and total, cross sectonal database 2006 Total Rotatonal group 1 Rotatonal group 2 Rotatonal group 3 Rotatonal group 4 Number % Number % Number % Number % Number % Total (DB120 = 11 to 23) Address contacted (DB120 = 11) Address non-contacted (DB120 = 21 to 23) Total address non-contacted (DB120 = 21 to 23) Address cannot be located (DB120= 21) Address unable to access (DB120 = 22) Address does not exsts or s nonresdental address or s unoccuped or not prncpal resdence (DB120 = 23) DB120=23 nclude also households where selected person ded or moved to nsttuon or abroad. Table 11: Dstrbuton of address contacted by household questonnare result and by household ntervew acceptance. Rotatonal group and total, cross sectonal database 2006 Total Rotatonal group 1 Rotatonal group 2 Rotatonal group 3 Rotatonal group 4 Number % Number % Number % Number % Number % Total Household questonnare completed (DB130 = 11) Intervew not completed (DB130 = 21 to 24) Total ntervew not completed (DB130 = 21 to 24) Refusal to co-operate (DB130 = 21) Entrely household temporarly away for duraton of feldwork (DB130 = 22) Household unable to respond (llness, ncapacty, etc.) (DB130 = 23) Other reasons (DB130 = 24) Household questonnare completed (DB135=1+2) Intervew accepted for data base (DB135 = 1) Intervew rejected (DB135 = 2) DB110 s not cross sectonal varable, so we calculated dstrbuton of DB110 from longtudnal database /56

28 Table 12: Dstrbuton of DB110 and DB010 n longtudnal database DB010=2005 DB010=2006 Total At the same address as last ntervew Entre household moved to prvate household wthn the country Entre hosehold moved to a collectve household or nsttuton wthn the country Household moved outsde the country Entre household ded Household does not contan sample person Address not contacted Splt-off household New address added to the sample ths wave or frst wave Fuson Total Table 13 Dstrbuton of DB110 and DB120 accordng to the DB010 n longtudnal database DB010=2005 Address contacted DB120=11 Address does not exst or s nonresdental address or s unoccuped or not prncpal address DB120=23 DB010=2006 Address contacted DB120=11 Address does not exst or s nonresdental address or s unoccuped or not prncpal address DB120=23 At the same address as last ntervew Entre household moved to prvate household wthn the country Entre hosehold moved to a collectve household or nsttuton wthn the country Household moved outsde the country Entre household ded Household does not contan sample person Address not contacted Splt-off household New address added to the sample ths wave or frst wave Fuson Total Accordng to the gudelnes for longtudnal componet we calculated RB120 only for cases where DB110=2,8,9. 28/56