Home

User Guide

1. Sociology Economics Political Science Public Health Statistics Methods Psychology Demography Geography Techn Science Psychiatry Theology Law Media Science Ethnology 1 7 Getting more information Questions Please write to swisspanel fors unil ch Phone 41 21 692 37 30 Fax 41 21 692 37 35 The following persons will be happy to answer Topics Information by Denise Bloch E mail and phone Registration data contract secretariat research net work conferences Data methods income and simulated taxes CNEF programming in STATA Data communication with the households instruction of interviewers monitoring of the survey Interviewer data contact data methods program ming in STATA Data methods communi cation with the households programming in SPSS and HLM Data programming in SPSS students question naires maintenance www swisspanel ch Project information ques tionnaires and documenta tion preparation and moni toring of the survey data dissemination including use of SHP data in a teach ing context Methods attrition analysis programming in SPSS and STATA Attrition analysis weight ing survey methodology programming in SAS Data data sets communal data programming in SPSS Ursina Kuhn Florence Lebert Oliver Lipps Val rie Anne Ryser Flurina Schmid Robin Tillmann Marieke Voorpostel Bryce W
2. The questions on income have changed over the duration of the panel cf Table 5 3 11 With the exception of family allowances only asked from 2004 onward as well as old age pension and other income sources in 1999 old age pension not asked in 1999 these changes should not influence comparisons across waves The variables collected from 1999 2001 can be constructed for all years by aggregating different income sources as shown in the table Table 5 3 11 Collection of individual income by wave 1999 2000 2001 2002 2003 from 2004 ISSWY ISSWY I EMPY I EMPY ISSINDY ISSINDY ISSAVSY ISSOASIY IS OASIY S AIY HAIY I PENY I PENY I STPY I STPY I UNEY IS UNEY ISSWELY ISSWELY IS GRAY IS GRAY ISSINSY ISSINSY _ 5 ISSFAMY ISS STFY I STFY I PIHY IS PIHY I PNHY I PNHY IS OSY IS OSY ISSOSY ISSOSY Household income There are two different ways of constructing household income Firstly in the household questionnaire reference persons are asked to estimate total household income sum of all household members Secondly in the individual questionnaire household members from 14 years of age are asked about their personal income Total individual income amounts corrected for within household transfers are then added to calculate house hold income The constructed variables on household income listed below represent the sum of individual income in two cases either if all individuals have answered the in come questions in the ind
3. 52 0 50 3 44 3 Mean satisfaction with health 0 10 8 23 8 20 8 17 sd 1 79 1 78 1 97 68 Participate in clubs 52 7 49 1 44 1 Mean participated in polls 0 10 7 93 7 37 7 18 sd 2 98 3 11 3 32 Mean general trust in people 0 10 5 54 5 47 4 96 sd 2 45 2 34 2 60 Mean interest in politics 0 10 5 83 5 29 4 97 sd 2 84 2 80 3 01 Mean political influence 0 10 3 58 3 81 3 45 sd 2 71 2 61 2 72 Mean trust in government 0 10 5 32 5 43 5 21 sd 2 18 2 21 2 33 Out of sample left the country institutionalized P Region Lake Geneva VD VS GE Middleland BE FR SO NE JU North west Switzerland BS BL AG Zurich East Switzerland GL SH AR Al SG GR TG Central Switzerland LU UR SZ OW NW ZG Ticino Only asked to respondents with the right to vote Asked from 2002 onwards 69
4. Learned professions LP Educational trajectory ED Work life WL Family events FE Retirement RE SaaS oS In order to assess the potentially negative impact of the self administered biographical questionnaire on the participation in subsequent waves of the yearly CATI a test sur vey was conducted in 2001 The results showed that the drop out rates did not increase substantially as a result of the questionnaire sent in between two waves Scherpenzeel et al 2002 Consequently the main survey was carried out in 2002 with those partici pants that had not been part of the test survey SHP_I biographical data are available for 5 560 individuals with the 2001 and 2002 sur veys combined Therefore some variables only exist for one of the survey years e g education history only for 2002 or only in an aggregated form e g living arrangement for 2001 The overall participation rate was 53 but over 80 among fully longitudinal The paper and pencil questionnaire is not available in English but only in the interview languages German French and Italian 38 panel survey respondents years 1999 2004 participated in the biography survey Bu dowski and Wernli 2004 The Biography Files include a two horizontal files with lines representing individuals Biography Master File SHPO_MBI and Biography Data File SHPO_BH_USER and b vertical files for each of the eight domains with lines representing events and
5. O F S Schweizer Haushalt Panel Panel suisse de m nages A swiss foundation for research Swiss Household Panel in social sciences Swiss Household Panel User Guide 1999 2008 Wave 10 December 2009 By Marieke Voorpostel Robin Tillmann Florence Lebert Bryce Weaver Ursina Kuhn Oliver Lipps Val rie Anne Ryser Flurina Schmid Boris Wernli Acknowledgements The Swiss Household Panel data are collected within the framework of the research program Living in Switzerland financed by the Swiss National Science Foundation The SHP is based at the Swiss Centre of Expertise in the Social Sciences FORS in Lausanne This guide is also based on the work of past members of the SHP Team How to cite this document Voorpostel M Tillmann R Lebert F Weaver B Kuhn U Lipps O Ryser V A Schmid F amp Wernli B 2009 Swiss Household Panel Userguide 1999 2008 Wave 10 December 2009 Lausanne FORS Correspondence to Swiss Household Panel FORS c o University of Lausanne B timent Vidy CH 1015 Lausanne swisspanel fors unil ch 2009 FORS User Guide SHP Table of contents Introduction 1 1 Aim 1 2 Organisation and funding 1 3 Use of the SHP 1 4 SHP and CNEF 1 5 Access to the data and data protection rules 1 6 Research network Living in Switzerland 1 7 Getting more information Study design 2 1 General design of the SHP 2 2 Sample design 2 2 1 Sampling plan 2 2 2 Sampli
6. and for each month you should tell me whether month would like you to tell month would like you to month would like you to tell your main activity was full time me if you have worked full tell me if you have worked me if you have worked full time employee part time employee full time or part time or if you full time or part time or part time or if you have not time self employed part time self have not worked due to a worked due to a period of un employed unemployed retired period of unemployment employment training or other training education housework or training or other reason reason any other situation 1 fulltime job gt 37h 1 1 fulltime paid job gt 37h 1 1 fulltime paid job gt 37h 1 1 Employee fulltime 1 2 part time job 19 36h 2 2 part time paid job 19 36h 2 2 part time paid job 19 36h 2 2 Employee part time 2 3 small part time job 1 18h 2 3 small part time job 1 18h 2 3 small part time job 1 18h 2 3 Self employed fulltime 1 4 unemployed 3 4 no job 5 4 unemployed 3 4 Self employed part time 2 5 continued education voca 4 5 continued education voca 4 5 Unemployed 3 tional retraining tional retraining 6 other 4 6 retired 4 6 Retired 4 7 other 4 7 Student 4 8 student 4 8 At home domestic work chil 4 dren 9 Other inactive 4 36 5 1 4 Last job file This file contains information on the last job of all individuals who were a inactive at the time of their first interview and b interviewed
7. s last occupation WRIGHT3 GOLDTHORPE ESeC CSP TREIMAN CAMSIS Table 5 3 7 Variables used for classifications for father s and mother s occupation profession education WRIGHT3 GOLDTHORPE ESeC CSP TREIMAN CAMSIS Variable name WR3LAJ GLDLAJ ESECLJ CSPLAJ TR1LAJ CAILAJ Variable name WA3FAJ WA3MOJS GLDFAJ GLDMAu ESECFA ESECMO CSPFAJ CSPMAJ TR1FAJ TR1MOJ CAIFAJ CAIMOJ profession education ISALAJSS ISALAJSS IS3LAJj P W111 IS4LAJ P W111 IS4FAJ IS4MOJ IS4FAJ IS4MOJ IS3FAJ IS3MOJ P 012 P 029 IS4FAJ IS4MOJ P 012 P 029 EDUCAT EDUCAT P 017 P 034 P 017 P 034 A The Wright class structure Wright III Hierarchical level man agement supervision production P W117 P W117 P W117 P W117 P W117 Hierarchical level man agement supervision roduction P 016 P 033 P 016 P 033 P 016 P 033 P 016 P 033 P 016 P 033 Number of employees of self employed P W114 P W114 P W114 P W114 P W114 status self employed employee etc P W112 P W112 P W112 P W112 P W112 gender SEX Number of status self employees of employed self employee employed etc P 014 P 013 P 031 P 030 P 014 P 013 P 031 P 030 P 014 P 013 P 031 P 030 P 014 P 013 P 031 P 030 P 014 P 013 P 031 P 030 The classifi
8. w10 2793 2718 6905 5740 4494 1127 2060 91 82 SHP II 2008 w5 1663 1546 3984 3291 2410 647 1400 81 80 64 Table B 3 Participation in the SILC Survey 2004 2005 SILC_1 11 Number of partici pating units Households with grids completed Household inter view completed Persons living in participating house holds Persons aged 14 years and older eli gible for individual interviewing Personal interview completed Proxy Interviews a Persons responding in current and all previous waves Grid level net re sponse rates b Individual level net response rates c SILC 2004 w1 2 287 2 199 5 450 4 441 3 498 73 90 SILC 1 2005 w2 1 103 1 064 2 689 2 168 1 683 79 83 SILC II 2005 w1 1 263 1 204 3 015 2 470 1 797 76 80 Source Swiss Household Panel 1999 2008 http www swisspanel ch The SHP participated in a joint venture project Living in Switzerland 2020 aimed at conducting the Statistics of Income and Living Conditions SILC pilot study 2004 2005 in collaboration with the Swiss Federal Statistical Office The SILC pilot data were distributed by the SHP until the end of 2008 65 Appendix C Attrition by demographic characteristics and social involvement Below demographic characteristics and social involvement attitudes and behaviour of both samples of the SHP are presented for respondents with di
9. 15 years old All individuals who were personally interviewed in any of the waves are included Unique information about a person s social origin is collected during the first interview It mainly relates to the composition of the household in which the person lived at the age of 15 and to the level of education and professional activities of both parents Persons who are not yet 20 years old and still living with their parents are not asked about their parents employment status Note that individuals who have had their first interview be fore they turned 20 are not in the social origin file Given the uniqueness of this information it doesn t make sense to attach it to each of the consecutive yearly waves Therefore the social origin module constitutes a specific file containing variable names in which the usual two digit number showing the year of the data collection is replaced by A separate variable OSYY indicates the wave during which the data on the person s social origin have been collected The questions corresponding to the variables P 060 to P 065 have only been asked in the first wave 1999 P 060 At age 15 Work in private households Employer Father P 061 At age 15 Public Company status Father 37 P 062 At age 15 Work in private households Employer mother P 063 At age 15 Public Company status Mother P 064 At age 15 Work in private households Employer Other person P 065 At age 15 Public Compa
10. 33 and in the preceding wave The activity calendar is empty for waves in which a respondent did not answer the indi vidual questionnaire The variable names in the calendar file are as follows JAN activity status in January in the year FEB activity status in February in the year MARS activity status in March in the year etc The calendar questions in the questionnaire have changed twice over the course of the years Three periods can be distinguished wave 2 amp 3 wave 4 amp 5 and wave 6 and thereafter For all waves however the professional status at the time of the survey is determined by the variables P W01 to P WO03 to distinguish between working for pay and not working for pay P W39 and P W42 to distinguish between fulltime and part time employ ment P W0O6 to distinguish between unemployment and inactivity The respondents who did not work during the week preceding the survey or did not have a job are asked the following question variable P W154 You are not currently in paid employment However since month year have you had a paid job also be it casual or on an irregular basis Respondents who worked at the time of the survey were asked the following question variable P W177 Since month year has there been a change in the number of hours you work have you started or ended an activity or even been unemployed wave 2 to wave 5 Since month year have you changed
11. Documentation Table 5 3 5 to 5 3 7 shows the variables used to construct the different classifications The classification of respondent s last job is4laj father s occupation and mother s occupation is done in the same way The following explanation of the construction of the classification for respondent s current occupation is therefore also applicable to respondent s last occupation and father s and mother s occu pation Table 5 3 5 Variables used to construct classifications for respondent s current ccupation Variable profession education Hierarchical Number of status gender name level man employees self agement of self employed supervision employed employee production etc WRIGHT3 WR3MAJ IS4MAJ S EDUCAT P W34 P W31 P W29 GOLDTHORPE GLDMAJ IS4MAJSS P W34 P W31 P W29 ESeC ESECMJ IS3MAJ P W34 P W31 P W29 CSP CSPMAJ P W28 EDUCAT P W34 P W31 P W29 TREIMAN TR1MAJ IS4MAJSS P W34 P W31 P W29 CAMSIS CAIMAJ P W28 SEX cf Joye and Schuler 1995 For a discussion on how occupations are to some extent reflections of their national and temporal context see Levy 2002 10 If some minor adjustments are made in order to adapt it to the European context the label ISCO 88 COM is used Cf International Labour Office 1990 International Standard Classification of Occupations ISCO 88 Geneva ILO 44 Table 5 3 6 Variables used to construct classifications for respondent
12. and consumed within the family unit obtained from the outside or provided by external bodies e g care for children and the elderly e Labour market participation work and life satisfaction What are the different forms of labour market participation full time vs part time employment precari ous and insecure employment sub employment vs over employment under and over qualification etc and their relationship to work and life satisfaction How do people especially women with small children manage conflicting de mands from the workplace and from home e Poverty and social exclusion What kinds of living conditions are associated with poverty and social exclusion What are the family and individual characteristics of the poor and what is the mechanism which leads them out of poverty Who remains poor despite policy measures for support What are the complex rela tionships between poverty social isolation and externally induced social exclu sion e Gender social and economic participation How do life trajectories diverge ac cording to gender Why do professional careers of men and women with similar educational resources still diverge e Social determinants of health How is the life course of individuals and families of widely different origins and social conditions related to health behaviour and out comes What are the consequences of worsening living conditions on health What impact does ill health have on living conditions
13. bias in only one year 20 out of 38 Since the bias only appears in a limited number of years this is not to discourage researchers from using these variables It is more to encourage them to use some caution in the use of these variables particularly those that are affected in many years Furthermore the 24 test only measures for attrition bias on population totals it is not known whether this has any effect on the interaction between variables such as coefficients in regressions Table 4 4 The list of variables and years with bias after weighting VAR99 CIVSTA99 OCCUPA99 P99A11 P99A13 P99A18 P99A19 P99A43 P99A53 P99A57 P99C01 P99C11 P99D29 P99E15 P99F23 P99101 P99N34 P99N35 P99N39 P99N40 P99N41 P99N42 P99N45 P99N46 P99P01 P99P04 P99P06 P99P07 P99P08 P99P09 P99P19 P99P24 P99P26 P99P28 P99P29 P99P30 P99P40 P99W04 P99W43 Label99 Civil status in year of interview Actual occupation from grid Leisure Do it yourself gardening Frequency Leisure Disco Frequency Leisure Theatre opera exhibition Frequency Leisure Cinema Frequency Free time Weekend Number of hours Leisure Gardening Frequency Leisure Cycle motorcycle car Frequency Health status Number of days affected by health problems Last 12 months Partner Yes no Education Current training Type Child care outside HH without payment Last 12 months Satisfaction with financial situation Participation
14. details about the last main job held These modules also comprise objective elements such as profession status of the profession the number of hours worked work schedule atypical work as well as subjective elements such as satisfaction with various aspects of the job the evaluation of promotion prospects or of personal qualifications 6 income including objective elements such as total personal income total profes sional income received social transfers received private transfers and other income and subjective elements such as satisfaction with the financial situation and an evalua tion of changes concerning the personal financial situation 7 participation integration networks taking into account objective elements such as frequency of social contacts non remunerated work outside home participation in asso ciations membership of and participation in religious groups and subjective elements 15 such as the assessment of social capital by means of evaluation of potential practical help and emotional support from various social networks 8 politics and values referring to objective elements such as political participation membership party identification political positioning and subjective elements such as satisfaction with the political system the evaluation of issues or even political values and finally 9 leisure and media comprising objective elements such as leisu
15. follows shortly after the initial mail approximately one week later the letters are sent in three mailings with an interim of one week Enclosed with the preliminary mail participants receive a newsletter containing some results of recent analyses of the SHP data for further information see 3 3 4 Households that did not respond since at least one wave are contacted at a later point in time also divided in three groups They are treated like households refusing in the cur rent wave Therefore these households are contacted by the best performing interview ers who received special refusal conversion training see also 3 3 3 3 2 Selection and training of interviewers and supervisors To guarantee smooth functioning of the fieldwork M I S Trend employs a large group of interviewers plus especially trained supervisors Before the start of the fieldwork inter viewers and supervisors participate in a training consisting of two sessions The supervisors training aims to prepare the supervisors for their roles as contact per sons organizers of the interviews and supervisors of the interviewers The supervisors who are experienced interviewers are responsible for the performance of the inter viewers The aim of the interviewers training is to become familiar with the SHP in general with its longitudinal design and the specific difficulties Complex items are discussed and the interviewers learn how to convince respondents t
16. not individuals if appropriate SHPO_BV amp amp _USER SHPO_MBI The Biography master file contains the identification numbers idpers of all individuals who answered the biographical questionnaire in 2001 or 2002 The master file further includes individual population weights wpOOtbgp and sample weights wpOOtbgs For methodological reasons weights of zero had to be attributed to 199 persons SHPO_BH_USER In the horizontal file each row represents one respondent It contains in total 281 vari ables representing for each domain per episode the beginning end and description For example for every employment starting date end date and several characteristics of the job are included all as separate variables Also individual population weights wpOOtbgp and sample weights wpO0tbgs are included in this file The vertical files 1 Living arrangements SHPO_BVLA_USER 2 Periods outside of Switzerland SHPO_BVSA_USER 3 Changes in civil status SHPO_BVCS_USER 4 Learned professions SHPO_BVLP_USER 5 Educational trajectory SHPO_BVED_USER 6 Work life SHPO_BVWL_USER 7 Family events SHPO_BVFE_USER 8 Retirement SHPO_BVRE_USER In the eight vertical files one file per domain a row represents one episode Re spondents experiencing different episodes in a given domain for example they have held several jobs take up multiple rows in the file one for every job An index variable is included to preserve the order of the e
17. quality of work measured objectively and subjectively Three forms of integration deviates from this model insecure integration int gration incertaine is the result of unstable job but good working conditions and satisfaction at work constrained integration int gration laborieuse is the product of a stable job but with work con straints leading to dissatisfaction and disqualifying integration int gration disqualifi ante corresponds to the combination of job instability and hard working conditions Paugam 2000 For more details see Bergman Lambert Prandy and Joye 2002 50 5 3 5 Income Respondents are asked about various income sources and total income both in the indi vidual and in the household questionnaire They are free to report gross or net amounts after deduction of social security contributions and to report monthly or annual income At the basis of these questions variables on yearly income amounts are constructed Both net and gross incomes are constructed using standard assumptions on social secu rity contribution see www swisspanel ch If monthly income has been indicated by re spondents annual income is calculated using information from two different sources the number of months the respondent has received this income on the one hand and the activity calendar on the other All constructed variables have passed a series of man ual plausibility checks These checks involve typing m
18. the end of 2008 During the whole period at the University of Neuchatel the SHP has contributed to academic teaching The third phase of the SHP is linked to the integration to the Swiss Centre of Expertise in the Social Sci ences FORS Since 2008 the SHP still funded by the Swiss National Science Founda tion is indeed located at the University of Lausanne within FORS 1 3 Use of the SHP When analysing research domains indicated by the SHP data users n 633 we found that 1137 topics of interest were mentioned Figure 1 shows the relative importance of the single topic categories given by the SHP research network members The categories Health Physical Activity leads the table but still represents only 9 of all domains of interest mentioned by the active members of the SHP Research Network Lifecourse Adolescence Retirement Aging Povery Living Conditions Quality of Life and Labour Market Employment Income each 9 represent almost as fre quently researched topics Most importantly the analysis shows that the active data users of the SHP research network cover a very broad spectrum of research domains This is a strong indication that the multidisciplinary SHP survey serves the research needs of a very diversified and interdisciplinary academic community both nationally and internationally Figure 1 1 References to different areas of interest by SHP data users Health Physical Activity EEE
19. the sample of households and individuals selected and inter viewed for the first time in 2004 In a household panel information is collected at various levels household individual so several questionnaires are used The SHP uses three types of questionnaires the household grid lasting less than 10 minutes the household questionnaire lasting 15 minutes on average the individual questionnaire including a proxy questionnaire for those who are absent for a long period who are handicapped too ill to respond or younger than 14 years All individuals aged 14 or more living in the household are eli gible to answer the individual questionnaire lasting around 35 minutes 2 2 Sample design The first sample SHP_I is a stratified random sample of private households whose members represent the non institutional resident population in Switzerland In 1999 the methodology section of the Swiss Federal Statistical Office drew a random sample in each of the seven major statistical regions of Switzerland on the basis of the Swiss tele phone directory SRH Stichprobenregister f r Haushalterhebungen which covered over 95 of all private households Selected in this way the households are a represen tative sample of the various social groups in all regions of Switzerland However as the interviews are carried out in the three official national languages German French and Italian only there might be a certain bias concerning how pop
20. your professional status employee self employed changed the amount of hours you work full time part time started or stopped work or been unemployed wave 6 and after In case the answer is no on one of these questions the activity status by the time of the interview is assumed to hold for every month that elapsed since the preceding inter view or for the last 12 months if the respondent did not respond to the individual ques tionnaire in the preceding wave For these cases the appropriate value is imputed for all months since the last wave In case the answer is yes on one of the questions above i e if the person reported any changes in his her status during the period considered the calendar questions are asked and the employment situation is assessed for every month since the previous wave The calendar questions changed twice since the start of the survey First in wave 2 and 3 different questions were asked depending on whether or not the respondent had a paid job Response categories differed between these two questions see Table 5 1 1 In wave 4 and 5 both active and inactive respondents answered the same questions in the calendar with slightly adapted response categories compared to earlier waves Up 34 to wave 5 it is possible to distinguish between large and small part time jobs From wave 6 onwards this distinction is no longer made but separate response categories for self employed respondents and e
21. A weight share is done on all households to give all members of the household the same weight The sharing is the same as equation 4 where P may be equal to 0 Next this weight is adjusted by segmentation for the non response to the household questionnaire to form PTM_NRQM This segmentation is done at the household level All individuals including those under 14 are then assigned this calculated household weight Finally a calibra tion is done under the restriction that all members of the same household must have the same weight The calibration is as above 4 2 4 Weights for the combination of the two panels These weights are designed for performing analysis longitudinal or transversal us ing both the SHP I and SHP II combined They are considered preferable as they pro vide the largest possible sample size The combination of the two panels occurs at the level of the seven regions The combination is a so called convex combination This kind of combination has the form of multiplying the weights for members of the SHP by a factor p and for members of the second panel by 1 p By doing this the introduc n L n n tion of new bias of sample sums is avoided The factor of combination is p where n is the number of responding units from the first panel and n is the number of responding units from the second panel The unit is either the person in the case of the individual weights or the household in the ca
22. Class Structure in Modern Britain Oxford Clarendon Press Goldthorpe J amp Hope K 1974 The social grading of occupations Oxford Clarendon Press Graf E 2008 Pond rations du PSM PSM_1 vague 8 PSM_II vague 3 PSM_1 et PSM_II combin s N 338 0054 Neuchatel Office F d ral de la Statistique Graf E 2009a Ponderations du Panel suisse de m nages PSM_1 vague 9 PSM II vague 4 PSM_1 et PSM_II combin s Nech tel Office F d ral de la Statistique Graf Eric 2009b Etude empirique de l attrition du Panel suisse de m nages Neucha tel Office f d ral de la statistique OFS Groves R M 2006 Nonresponse rates and nonresponse bias in household surveys Public Opinion Quarterly 70 646 675 Groves R M amp Peytcheva E 2008 The impact of nonresponse rates on nonre sponse bias Public Opinion Quarterly 72 167 189 Haferkamp H 1990 Sozialstruktur und Kultur Frankfurt am Main Suhrkamp H pflinger F Charles M amp Debrunner A 1991 Familienleben und Berufsarbeit Zum Wechselverh ltnis zweier Lebensbereiche Z rich Seismo International Labour Office 1990 International Standard Classification of Occupations ISCO 88 Geneva ILO Joye D and Scherpenzeel A 1997 Observation a long terme Projet de panel Pro gramme Prioritaire Demain la Suisse Bern Fonds national suisse de la recherche scien tifique Joye D amp Schuler M 1995 Stratification s
23. F So far 16 special contracts for CNEF Data have been signed for the SHP Figure 3 shows the continuous increase of SHP data users since the first wave Figure 1 3 Users ever received a SHP CD Final version SHP Data 700 600 500 400 300 200 100 0 wave 1 wave 2 wave 3 wave 4 wave 5 wave6 wave 7 wave 8 wave 9 Moreover the prevalence of the various disciplines among the SHP data users n 633 is shown in Figure 4 Among the SHP data users sociology 42 and economics 22 are by far the most prevalent disciplines followed by political science 10 public health 6 statistics 6 and psychology 3 But a few technical scientists and ge ographers are also present indicating that spatially related topics are also being ana lysed using the SHP data The data users belong to the following institutions Swiss academic institutions 66 International academic institutions 17 Public administrations 10 and Private insti tutes 6 National and international academic communities clearly dominate compris ing 83 of all data users but the statistical use by public administrations and private research facilities is certainly not negligible Within Switzerland all universities and many universities of applied science Fachhochschule HES are represented among the data users Figure 1 4 Disciplines and their distributions among SHP data users CD from any wave n 607
24. G Ticino TI Uri UR Valais VS Vaud VD Zug ZG Zurich ZH 62 Appendix B Participation in the Swiss Household Panel Table B 1 Participation in the Living in Switzerland Panel Survey 1999 2008 SHP_I Number of partici SHP SHP _I SHP SHP _I SHP _l SHP SHP _l SHP _l SHP _l SHP pating units 1999 w1 2000 w2 2001 w3 2002 w4 2003 w5 2004 w6 2005 w7 2006 w8 2007 w9 2008 w10 Households with 5 074 4 532 4 314 3 685 3 289 2 918 2 526 2 580 2 893 2793 grids completed Household inter 5 074 4 425 4 139 3 582 3 227 2 837 2 457 2 537 2 817 2718 view completed Persons living in 12 931 11 678 11 116 9 537 8 478 7 517 6 491 6 587 7 225 6905 participating house holds Persons aged 14 10 293 9 297 8 942 7 553 6 719 5 976 5 220 5 333 5 972 5740 years and older eli gible for individual interviewing Personal interview 7 799 7 073 6 601 5 700 5 220 4 413 3 888 4 091 4 630 4494 completed Proxy Interviews a 2 638 2 381 2 174 1 984 1 724 1 482 1 241 1 237 1 226 1127 Persons responding 6 335 5 429 4 480 3 888 3 076 2 622 2 399 2 209 2060 in current and all previous waves Grid level net re 64 91 88 86 90 82 91 87 86 91 sponse rates b Individual level net 85 84 88 89 88 85 87 81 81 82 response rates c Source Swiss Household Panel 1999 2008 http www swisspanel ch a The SHP proxy interviews include information about children under 14 years and adult persons unable
25. Se 106 Lifecourse Adolescence Retirement Aging EEE EEE 102 Poverty Living Conditions Quality of Life 99 Labour Market Employment Income EEE BEE EEE SRH 98 Survey Methodology Longitudinal Analysis Education Social Capital Culture Macro Economy Economic Policy Social Justice Social Security Family Household Composition Tasks Working Conditions Work Life Arrangements Democracy Party Political Behaviour Migration Minorities Mobility Gender Lifestyle Leisure Internet Marketing Social Stratification Values Religions Social Participation Networks Social Support Environment Housing Housing Market Region Ind Behav Consumption Emotions Coping Deviance Victimisation Social Reporting 100 120 Since the inception of the SHP in 1999 a great variety of issues of social and economic significance have been continue to be or can be studied using the SHP data For ex ample e Evolving patterns in changing living conditions quality of life and life satisfaction Who is progressively better or worse off and why What are the necessary living conditions for warranting a good quality of life Which objective and subjective factors most strongly determine life satisfaction e Family life and interaction with society at large What are the consequences of various forms of living together in terms of social support and solidarity Which services are produced
26. a future wave However 1 We permanently drop for following waves households that were not contacted at all during the 1st wave or those that did not supply any information at the time of the 1st wave not even a grid or those who only completed a non response questionnaire for wave 1 2 For SHP_I we permanently dropped all households that only replied to the grid at wave 1 For SHP_II we changed this rule and only permanently rejected households that had only completed a grid for wave 1 if at wave 2 we did not collect more than its grid 3 We drop households who give a final refusal households where no one is willing to respond to a household interview after refusal conversion attempts those who move away from Switzerland and those who are fully and permanently institutionalized 13 2 3 2 Initial rules governing the follow up of individuals 4 Respondents OSM Original Sample Member and their children are continuously followed whereas cohabitants are only re interviewed as long as they live with an OSM From wave 9 onwards also cohabitants are followed 5 The minimum age of eligibility is 14 years 6 As a general rule respondents OSM are followed until they die or are permanently institutionalized or leave the target population for another reason 7 Individuals giving a final refusal for two consecutive waves as well as those who send us a written refusal are rejected 2 3 3 Additional rules to fight
27. ace The individual master file SHP_MP contains all individuals who have resided in the par ticipating households in any of the waves This file includes gender date of birth month and year identification number of father and mother response status and interview dates for all waves 5 1 2 Annual files households and individuals The annual household files SHP99_H_ USER SHPOO_H_USER etc contain informa tion household interviews complemented by certain information from the grid question naire For the constructed variables see 5 3 The information from the yearly individual interviews SHP99_P_USER SHPOO_P_USER etc is included in the annual individual files For the constructed variables in these files see 5 3 For the complete questionnaires see Questionnaires under Documentation on www swisspanel ch 5 1 3 Calendar file Using the answers in the individual questionnaire the calendar file contains for every person the activity status in each month If the person has answered the individual questionnaire in wave x information on his her activity is contained for the past year last 12 months if the person has not answered the individual questionnaire in the preceding wave the period between the individual interview in wave x 1 and the individual inter view in wave x if the person has answered the individual interview both in wave x 2 In terms of labour market situation Here the term activity is used
28. ach of these groups is the factor of correction for the weight in ques tion transversal longitudinal etc for the members within each group If non response is random within the class the response bias after correction is minimal The selection of the RHGs is an iterative process It begins with one class that is the entire dataset and a list of dummy variables Then within a given class the variable that is determined to have the greatest effect on the likelihood of responding is then used to divide the class into two classes Although there are other choices for the manner of selecting of the variable the one used for the SHP is the algorithm CHAID which selects the variable that has the highest chi squared of Pearson This process repeats until each class satisfies one of the following conditions 1 none of the remaining variables is found to be significant on response rate 2 the number of members of the class would fall below a given level 50 in the case of the SHP and 30 in the case of the SHP II if the class were divided and 3 the response rate would fall below a given level 3 in the case of the SHP and 5 in the case of the SHP II if the class were divided 4 2 1b The calibrations After the adjustment for non response the weights are calibrated to four known popula tion estimates to get estimates that are representative of the Swiss population e sex age category 0 13 14 24 25 34 35 44 45 54 55 e th
29. aken to convert households who were abandoned earlier because of double refusals 4 1 2 Attrition As nonresponse can cause nonresponse bias in survey estimates Behr Bellgardt and Rendtel 2005 Groves 2006 Groves and Peytcheva 2008 there is a need for research on the impact of attrition on the representativeness of the longitudinal sample of the SHP and how this might impact variables of interest and research findings Analyses were performed to shed light on the impact of attrition in the SHP focusing on potential bias in 22 variables of interest Below a study on nonresponse and bias in variables of interest is reported We refer to Appendix C for a general impression of how respondents with different re sponse patterns differ from each other on demographic characteristics and several meas ures of social involvement A comparison is made between respondents who are in the panel every wave respondents with an irregular response pattern and respondents who have dropped out of the panel Note that calculations are based on unweighted data For the complete study we refer to the SHP Working Paper Voorpostel 2009 on the website www swisspanel ch A comparable study on attrition in relation to income as well as other studies on attrition can be found here as well Kuhn 2009 Lipps 2006a 2006b 2006c Voorpostel 2009 Attrition in variables of interest This is a brief discussion on the variables touched by attrition By touche
30. al and cultural groupings and c similar across all complex modern societies The Treiman Prestige Scale differs from Wright and Goldthorpe s class schema not only in that it measures subjectively attributed prestige as an indicator of ac cess to structural and functional power but also because it explicitly models a prestige hierarchy The prestige scores range between of 0 lowest prestige and 100 highest prestige Treiman 1977 49 F The Cambridge Social Interaction and Stratification Scale Camsis The Cambridge Social Interaction and Stratification Scale CAMSIS is based on the idea that social structure can be expressed by the social distance between individuals for instance through the co occurrence of occupations that individuals hold and the rela tionships that they form with each other CAMSIS has been developed initially from friendship networks and subsequently from cohabiting couples Stewart Prandy amp Blackburn 1980 For Switzerland the Population Census of 1990 was used to examine the probability of co occurrence of occupational titles between cohabiting couples In the simplest model the distances between occupations of couples are calculated on the basis of the contribution of the cell toward the x of a contingency table The x con tribution for each cell is entered into a traditional correspondence analysis which repre sents the best possible solution in a two dimensional space The first dimension
31. ances annual amount Might additionally be included in income from employment I PNHY Payments received from individuals not in household annual amount I PIHY Payments received from individuals in household annual amount I OSY Other income annual amount Might include 3 pillar inheritance income from capital such as income from wealth letting sub letting I PTOT I PTOTG gross Yearly total personal income annual amount I PTOTN net social security contributions on employment income de In most cases total income has been calculated by adding the different income sources In case of non response in any of the income sources and in some other cases in waves 1 to 5 total 51 ducted income refers to a global assessment of income Amounts of income sources which represent one off payments over 12 000 CHF are not considered in total income ISSWY I WYG gross Income from employment annual amount ISSWYN net social secu Takes account of 13 and 14 month salary bonuses or gratifi rity contributions de cations if applicable see www swisspanel ch ducted From 2002 on sum of I EMPY I INDY I STPY Social public transfers annual amount From 2002 on sum of I UNEY I WELY I GRAY I INSY IS STFY Income from private persons informal transfers annual amount From 2002 on sum of I PNHY I PIHY ISSAVSY Income from old age or disability pension annual amount From 2002 on sum of I OASIY ISSAIY ISSPENY
32. anguage of the variable and value labels Variables and values labels are available for each data file in French German Italian and English The files containing the syntax are Variable labels SHP_ WAVE QUEST _ LANGUAGE txt Value labels SHP_ WAVES QUESTS LANGUAGES txt WAVES is to be replaced by W1 Wave 1 W2 Wave 2 W3 Wave 3 W4 Wave 4 W5 Wave 5 W6 Wave 6 W7 Wave 7 W8 Wave 8 W9 Wave 9 W10 Wave 10 WA Waves ALL modules CA LJ MP MH OS QUESTS is to be replaced by 56 P Individual H Household X Proxy CA Activities calendar LJ Last Job MP Individual Masterfile MH Household Masterfile OS Social Origin LANGUAGES is to be replaced by E English F Fran ais D Deutch Italiano For SPSS labels To label a SPSS data file open the files located in the LABELS SPSS WAVES LANGUAGES directory in a syntax editor and run the syn tax For STATA labels To label a STATA data file open the files located in the LABELS STATA WAVES LANGUAGES directory in a do file editor and run the syn tax Note that all STATA file names variable names use lower case letters 97 References Beck U 1986 Risikogesellschaft Auf dem Weg in eine andere Moderne Frankfurt am Main Suhrkamp Verlag Behr A Bellgardt E amp Rendtel U 2005 Extent and determinants of panel attrition in the European Community Household Panel European Sociological Revi
33. as taken from the SRH 02 99 the SHP II from the SRH 02 04 The size of strata at the moment of the selection for SHP_I and SHP_II were as follows Table 2 2 Sizes SRH of strata at the moment of both selections Strata SHP_I N SHP_II N Lake Geneva region 7147725 648 590 Mittelland 837452 784266 North west Switzerland 484667 455 833 Zurich 646 469 587 850 Eastern Switzerland 5317731 493 606 Central Switzerland 313548 306605 Ticino 180623 1607123 Total 3 709 215 3436873 2 2 2 Sampling frame The population covered by the SHP consists of all individuals living in private house holds in Switzerland who had a telephone connection landline or mobile registered in the telephone directory Individuals living in old peoples homes institutions collective households or prison are not part of this population Persons who could not be con tacted by telephone or whose telephone connection is not listed are not covered by the survey undercoverage An estimated 98 5 of private households had a telephone connection at the time of the selection of the sample for the SHP_II in 2004 The SRH survey frame for household 12 surveys covered about 93 of these households In 1999 at the time of the selection of the sample for the SHP_I the SRH s coverage rate was about 95 The survey frame SRH is subject to the following errors e undercoverage some households were not listed in the directory at the time of selection This
34. as the age sex relations between the members of the household nationality level of education and occupational status 2 accommodation containing objective elements such as the type and size of the ac commodation home ownership or tenancy the cost of and or the subsidies received for housing as well as subjective elements such as satisfaction with the accommodation evaluation of the state of the accommodation and assessment of perceived nuisances 2 Out of the 1 520 asked again SHP_ 1 households in 2006 and 2007 580 completed at least the grid interview 14 3 standard of living referring to a list of goods owned by the household or activities that its members can carry out together with the reason financial or otherwise why goods are not owned or activities not carried out 4 the household s financial situation containing objective information such as the exis tence of financial difficulties and the household s reactions to different situations in debtedness and the reasons for it the total household income the amount of tax paid and the social and private transfers as well as subjective elements such as satisfac tion an estimate of the minimum income the household considers necessary or an evaluation of how the household s financial situation has evolved 5 the household and the family collecting information on any external help available to the household for housework or child ca
35. ations between educational attainment and occupational titles ii According to educational and training trajectories normally followed by those with a particular occupation as established from the Swiss Popula tion Census of 1990 iii Based on the respondents attained educational and professional qualifi cations whatever the relevance to their occupation Technically the following rules were followed a Owners Employers self employed and at least 10 employees b Petty bourgeoisie self employed and less than 10 employees c Managers Experts professional leading10 or supervisory role as well as an advanced educational attainment d Managers salaried with supervisory position and not yet classified in any of the above categories e Professionals salaried with advanced educational attainment but without su pervisory functions f Semi Professionnals salaried with either advanced or middling educational at tainment and with middling professional requirements g Worker other workers B Erikson Goldthorpe and Portocarero s Comparative Analysis of Social Mobility in In dustrial Nations schema CASMIN The first Goldthorpe class schema was based on occupation and occupational status self employed salaried Originating from Goldthorpe and Hope s prestige scale 1974 and Goldthorpe s subsequent class schema 1987 two levels of classification were de veloped that inclu
36. attrition The following measures were taken to reduce attrition rates from 2006 to 2008 waves 8 to 10 of SHP_I and waves 3 to 5 of SHP_II contacts with all SHP_I households which had refused to participate between 2000 and 2003 that is at waves 2 3 4 and 5 con tact with past final refusal households which participated again after renewal of contact in 2006 and 2007 contact with refusing SHP_I households in 2006 and 2007 waves 8 and 9 contact with refusing SHP_II households in 2005 2006 and 2007 waves 2 3 and 4 of SHP_II follow up of non original sample members 2 4 Questionnaires 2 4 1 Content of the questionnaires The Living in Switzerland survey is a comprehensive survey The questionnaires household and individual cover a broad range of social fields and topics They are also designed to collect both objective resources social position participation etc and subjective data satisfaction values evaluation etc The whole constitutes an opera tionalisation of the different elements of the microsocial level living conditions life events attitudes and perceptions and lifestyles ways of life Budowski et al 1998 A household panel collects data at two levels the household and the individual In the case of the SHP survey the questionnaire at the household level covers the following areas 1 composition of the household containing basic information about all the members of the household such
37. blications to the SHP swisspanel fors unil ch The following conditions regarding data use and data protection are to be strictly re spected by the authorized data users The use of the SHP data for commercial purposes is strictly forbidden e The SHP data are for the exclusive use of the person s signing the contract it is strictly forbidden to make part or all of the data available to a third party In a re search team all users have to sign the contract individually e Research topics and changes thereof are to be reported to the SHP team In tended publications or semi formal reports grey literature using the SHP data must be announced as soon as possible to the SHP team e A copy of research reports and manuscripts submitted to journals must be sent to the SHP Team In case of agreement of the author these documents will be made available on the SHP website e Users of SHP data automatically become members of the research network Liv ing in Switzerland contact information research interests and title of reports and publications are displayed on the SHP website e Publications in paper or electronic format based on SHP data must carry the fol lowing mention This study has been realized using the data collected in the Living in Switzerland project conducted by the Swiss Household Panel SHP which is based at the Swiss Centre of Expertise in the Social Sciences FORS University of Lausanne The project is financed
38. by the Swiss National Science Foundation see acquiring the data on our website for the French and German versions of this mention e Publications tables and figures based on SHP data in paper or electronic format must carry the following mention Source Swiss Household Panel SHP e Publications tables and figures based on SHP data in paper or electronic format must carry the following mention Source Swiss Household Panel SHP e Data users and collaborators are to comply with the Swiss Regulations regarding data protection in particular users are to refrain from trying to identify a particu lar household or person and are strictly prohibited to use information for anything other than scientific research purposes Published results shall not contain any information making it possible to identify any SHP household or interviewed per son 1 6 Research network Living in Switzerland In December 2009 the research network Living in Switzerland had some 633 regis tered members i e researchers who have signed a data contract since 1999 and re ceived at least once a SHP CD analysing the SHP data on a variety of topics types of households and families poverty health living conditions of elderly people living condi tions of first and second generation immigrants political participation and life satisfac tion etc Since spring 2008 SHP Data is also distributed as part of the Cross National Equivalent File CNE
39. cation presented here was developed several years after the first and second versions cf Western amp Wright 1994 It was used in particular for the study of social mobility Its main advantage already present in Wright s second classification rests in the explicit use of three dimensions authority expertise and property These dimen sions form seven categories instead of the twelve that Wright proposed in his second version The reduction from twelve to only seven cells obviously increases the cell counts and thus statistical power A number of choices were made for the operationalisation and adaptation of this schema some of which are necessarily rather arbitrary 11 This recodification differs slightly from that of Levy et al 1997 45 a Most cases of self employment were unproblematic In some cases we attrib uted this status to family members employed in their own family business as well as to those who considered themselves employees of their own enterprise b The demarcation between middle class and the petty bourgeoisie is often based on whether or not the respondent has employees Here by homogeneity with other classification schemas we set the minimum qualification criteria to ten employees c Competence derived from educational attainment are qualified in several ways i Directly relating to the occupation ISCO 88 includes in its occupational classification an explicit reflection on the rel
40. ce respondents to categorise their reaction towards an attitude object instead of directly mapping it onto the response continuum thus causing information loss Early research has already shown that respondents dif ferentiate more between objects when offered response scales with greater numbers of categories Bendig 1954 Garner 1960 The larger the number of points the more powerful the scale is in discriminating but at a certain point respondents become unable to make fine distinctions and thus round off 2 Improvement of data analysis Improving the measurement procedures is one way to improve the quality of data analy sis In their investigation of the possibilities to optimise measurement procedures in so Dr Be This section is a summary See www swisspanel under Documentation for the complete ver sion 16 cial science Van Doorn Saris and Lodge 1983 did not simply enlarge the number of scale points but used psychophysical scaling see also Lodge 1981 Respondents ex pressed their answers on continuous scales by drawing lines or assigning numbers to their opinions thus creating interval level measures The best alternative to category scales within the class of magnitude estimation scales that can be used in CATI is the number production scales It is essential that a magnitude estimation scale has fixed anchors or reference points The 11 point number scale used in the panel questionnaire has for example two re
41. d ual interview SHP_II With regard to the SHP_II waves 1 to 5 2 538 households and 3 654 individuals were first interviewed in 2004 In the fifth wave 1 546 households and 2 410 persons were answering At the household level see Table 4 1 A the drop in participation was highly significant in the second wave 29 as compared to 5 to 8 for the three other waves in the 2004 2007 period In 2008 the number of households validly interviewed increases due to renewal of contacts with households who where abandoned earlier be cause of double refusal like for the SHP_I At the individual level see Table 4 2 A the drop in participation was as at the household level particularly significant in the second wave 28 compared to the other waves in the 2004 2007 period when the drop in participation lay between 2 and 6 In 2008 the number of persons validly interviewed increases slightly due to 1 re Contrary to the SHP_I starting in 1999 the household recruited in 2004 were not explicitly asked to commit themselves for several years According to the interviewers many households were surprised to be called one year later to be interviewed again in the ongoing panel study 21 newal of contacts with past refusal households and 2 efforts made by the interviewers of M I S Trend to enroll all eligible household members for an individual interview However it should be noted that drop in participation is quite simi
42. d by attrition we mean the variables that demonstrate a bias after subsequent waves The technique used to detect bias is the one proposed in Graf 2009b It tries to capture the effect of attrition alone and not the effect of evolution due to time The technique is also not de signed to detect any bias that is introduced in the sample selection or non response in the first wave of the survey The test described in the section technique was done on the first panel SHP I only tests will be performed to see how the combined panels evolved and be released in a later report However this should give an idea of the vari ables that are related to attrition Not only is bias due to attrition searched for but it is checked whether the longitudinal weights correct for it A list of variables whose bias persists after weighting is listed in the section results These results should be consid ered preliminary as a more complete report will be released in 2010 Technique In order to assess the effect of attrition only individuals that responded in wave 1 1999 are considered The following subsets are therefore constructed R99 99 sL sL R99 99 00 sLo SL OSL VE sL risk sL sL AsL where sL is the set of respondents in 1999 and sL are the longitudinal respondents in year 20 The 1999 responses are then compared on these subsets even for the re spondents of later waves By using the 1999 responses only
43. d to the minor groups i e 3 digit groups of EU variant of the In ternational Standard Classification of Occupations 1988 ISCO88 COM e details of employment status i e whether an employer self employed or em ployee number of employees at the workplace whether a worker is a supervisor 12 See Bergman and Joye 2001 for a more detailed discussion 13 This classification was developed by a consortium of nine institutes from the UK Germany France the Netherlands Sweden Italy and Ireland See for more information http www iser essex ac uk research esec 47 Table 5 3 8 The European Socio economic Classification ESeC Class Common Term 1 Large employers higher grade professional Higher salariat administrative and managerial occupations 2 Lower grade professional administrative Lower salariat and managerial occupations and higher grade technician and supervisory occupa tions 3 Intermediate occupations Higher grade white collar workers 4 Small employer and self employed occupa Petit bourgeoisie or independents tions excluding agriculture etc 5 Self employed occupations agriculture etc Petit bourgeoisie or independents 6 Lower supervisory and lower technician oc Higher grade blue collar workers cupations 7 Lower services sales and clerical occupa Lower grade white collar workers tions 8 Lower technical occupations Skilled workers 9 Routine occupations Semi and nonskilled workers 10 Never worked and long ter
44. dar file Last job file Social origin file Biographical files Interviewer files 5 2 Variable naming conventions 5 3 Constructed variables 5 3 1 5 3 2 5 3 3 5 3 4 5 3 5 5 3 6 Socio demographic variables Education Work status occupation and social position Professional integration Income Geographical information 5 4 Missing value conventions 5 5 Imputation procedures 5 6 Combining data files 5 7 Changing the language of the variable and value labels References Appendix A List of Swiss cantons Appendix B Participation in the Swiss Household Panel Appendix C Attrition by demographic characteristics and social involvement 62 66 CHAPTER 1 INTRODUCTION 1 1 Aim The principal aim of the Swiss Household Panel SHP is to observe social change in particular the dynamics of changing living conditions and representations in the popula tion of Switzerland During the years 1998 2007 the SHP was a joint project run by the Swiss National Science Foundation the Swiss Federal Statistical Office and the Univer sity of Neuchatel Since January 2008 the SHP is part of the Swiss Centre of Expertise in the Social Sciences FORS located at the University of Lausanne The creation of the SHP was one of the key structural measures implemented by the Swiss Priority Program Switzerland Towards the Future during the period 1998 2003 for the following two main purposes Farago 1996 Joye and Scherpenze
45. ded seven or 36 categories Further development in conjunction with the CASMIN Comparative Analysis of Social Mobility in Industrial Countries project makes the seven category schema more suitable for comparative investigations and it has established itself as the most prominent schema for comparative intergenerational mobility studies Contrary to earlier versions the current incarnation requires information on the respondents number of employees and supervisory function As a class schema that is primarily used in comparative research it is most frequently based on ISCO 88 Ganzeboom and Treiman 2003 have successfully adapted the most recent Goldthorpe class schema and use the following codes 1 Higher controllers 2 Lower controllers 3 Routine non manual employees 46 4 Self employed with employees 5 Self employed without employees 7 Manual supervisor 8 Skilled manual employees 9 Semi and unskilled manual employees 10 Farm labour 11 Self employed farmers It is more difficult than with other schemas presented here to assess how respondents are classified because several dimensions are integrated in complex and unspecified ways C The European Socio economic Classification ESeC The European Socio economic Classification ESeC is a European occupationally based classification based on the Erikson Goldthorpe Portocarero EGP Schema The information required to create ESeC is e occupation code
46. derland NV KPN Research Scherpenzeel A Zimmermann E Budowski M Tillmann R Wernli B amp Gabad inho A 2002 Experimental pre test of the biographical questionnaire SHP Working Paper 5 02 Neuchatel Swiss Household Panel Schuler M Dessemontet P amp Joye D 2005 Die Raumgliederungen der Schweiz Neuchatel Bundesamt fur Statistik Schuman H amp Presser S 1981 Questions and answers in attitude surveys Experi ments on question form wording and context New York Academic Press Stewart A Prandy K amp Blackburn R M 1980 Social Stratification and Occupations London Macmillan Treiman D J 1977 Occupational prestige in comparative perspective New York Academic Press Van Doorn L Saris W E amp Lodge M 1983 Discrete or continuous measurement What difference does it make Kwantitatieve Methoden 10 104 120 Voorpostel M 2009 Attrition in the Swiss Household Panel by demographic character istics and levels of social involvement SHP Working Paper 1 09 Lausanne Swiss Household Panel 61 Appendix A List of cantons in Switzerland Aargau AG Appenzell Ausserrhoden AR Appenzell Innerrrhoden Al Basel Stadt BS Basel Landschaft BL Bern BE Fribourg FR Geneva GE Glarus GL Graubunden GR Jura JU Lucerne LU Neuchatel NE Nidwalden NW Obwalden OW Schaffhausen SH Schwyz SZ Solothurn SO St Gallen SG Thurgau T
47. done The baseline weights are described in some detail as they are the basis of all wave specific weights We then address the fashion that the individual longitudinal weights the individual transversal weights and the house hold transversal weights are constructed for the two panels independently A discussion on how the weights from the two independent panels are combined to form the three de livered weights for the combined panel is given Finally we close with some assistance in selecting and using the weights 4 2 1 Overview of techniques These are the principle techniques used in the formation of weights in the SHP context The process of segmentation is used to determine the probability of being in the panel The inverse of which is the basis of the weights Calibrations are then used to adjust the weights so that certain population sums are correct equal to the sums of the non institutionalized Swiss population The adjustments due to calibration are chosen to be as small as possible so that the introduction of bias for non correlated variables is mini mized 4 2 1a The process of segmentation Aside from the first wave weights of the SHP I modeling of non response in the SHP is done by the process of segmentation The goal of segmentation is to determine the re sponse probability for the panel members or households 26 The dataset is divided into response homogeneity groups RHGs The inverse of the response rate in e
48. e CNEF The CNEF con tains equivalently defined variables for the US Panel Study of Income Dynamics PSID the German Socio Economic Panel GSOEP the British Household Panel Study BHPS the Household Income and Labour Dynamics in Australia HILDA the Canadian Survey of Labour and Income Dynamics SLID the Swiss Household Panel SHP and the Korean Labor and Income Panel Study KLIPS The CNEF is designed to allow cross national researchers access to simplified versions of these panels For acquiring the data see http www swisspanel ch doc PSM CNEF index php lang en For more information see www human cornell edu che PAM Research Centers Programs German Panel cnef cfm or Frick et al 2007 1 5 Access to the data and data protection rules The SHP data are available at no charge Users must sign a contract available on our website http www swisspanel ch shpdata contract php lang en amp pid 23 Once the contract is signed users will have access to the most recent SHP Data The SHP data are available to researchers signing in person the data contract at no charge and exclusively for non commercial use It is strictly forbidden to attempt to iden tify particular households or individuals and to make parts or all of the data available to a third party In a research team all users are to sign the contract individually SHP data users commit themselves to personally send a copy of all working papers final reports or pu
49. e Swiss Household Panel SHP The general basis of the weight of an observation is the inverse of the estimated probability of responding If the response probability were perfectly known this would provide unbiased population totals for the variables of interest This discussion is de signed to give an idea of how the weights are produced and what techniques are used If one is interested in a detailed exposition on the production of the weights for a given year one should examine the documentation at http www swisspanel ch doc weighting php lang en amp pid 8 There are nine different weights produced for the SHP They are the individual longitu dinal individual transversal and the household transversal weights for the SHP I SHP Il and the combined SHP and II Of these 9 weights 4 are delivered in two forms each discussed below with the standard data sets 1 the individual longitudinal weights for the SHP 2 the individual longitudinal weights for the combined panel SHP I and SHP Il 3 the individual transversal weights for the combined panels and 4 the household transversal weights for the combined panel These four weights are selected because they provide maximum sample sizes for most studies Refer to 4 2 5 below for the selection of the appropriate weights according to the type of analysis The topics start with some of the technical points namely the process of segmentation and the fashion that calibrations are
50. e based off of the initial weight POIDINIT which is the weight at base line 1999 for SHP_I and 2004 for SHP_lI as described above In a given wave longitu dinal individuals who are the original sample members are modeled for response at the 28 level of the grid Only the adults aged 14 years and over are used The method of mod eling non response is segmentation Graf 2008 The inverse of the grid response rate for each RHG is the adjustment factor for the weight Specifically this adjusted weight is PURE PPT 3 T RHG where 7p is the response rate for the given RHG This adjusted weight becomes the basis for all of the wave specific weights The three types of weights presented below are all determined using the same method ology which combines segmentation and calibration using population characteristics 4 2 3a Individual longitudinal weights These weights are for conducting longitudinal analysis and provide estimates represen tative of the baseline population 1999 for SHP and 2004 for SHP Il Here the seg mentation is done on the response to the individual questionnaire for longitudinal indi viduals conditional on having responded to the grid no individuals are questioned be fore the grid is completed First a basic longitudinal weight PL_NRQI is produced from P_NRGRIL in the same way as documented by equation 3 Second to produce the wave specific longitudinal weight PL_NRQI is calibrated to estimated po
51. e number of individuals living in the regions l manique central Tessin com bined versus the number of individuals living in the regions mittelland north west Zurich oriental combined the number of individuals with Swiss nationality and e the number of individuals married One should note that values for age 0 13 are used only for the household transversal weights and that the number of individuals married is not available for the longitudinal weight for SHP I The calibration is done by minimizing a distance function between the weights adjusted for non response and all possible weights that satisfy the calibration estimates The macro was developed by Jean Fran ois Naud and Caroline Cauchon of Statistics Can ada 2006 The distance function used is the so called linear 2 G d w md where for individual k d is the weight before calibration and w is 2d 4k the weight after and q is a factor independent of d that is imposed to be 1 in the case of the SHP One problem with this distance function is that negative weights can appear As we consider negative weights to have little fundamental sense we do not allow such weights in the final output The method used for dealing with such weights when they do appear is to assign them their pre calibration weight and re calibrate 4 2 2 Formation of initial weights or baseline weights The initial weights are the basis of all wave specific weights They begin with a design
52. eaver Boris Wernli swisspanel fors unil ch 41 0 21 692 3730 ursina kuhn fors unil ch 41 0 21 692 3722 florence lebert fors unil ch 41 0 21 692 3715 oliver lipps fors unil ch 41 0 21 692 3724 valerie anne ryser fors unil ch 41 0 21 692 3740 flurina schmid fors unil ch 41 0 21 692 3716 robin tillmann fors unil ch 41 0 21 692 3721 marieke voorpostel fors unil ch 41 0 21 692 3727 bryce weaver fors unil ch 41 0 21 692 3746 boris wernli fors unil ch 41 0 21 692 3723 10 CHAPTER 2 STUDY DESIGN 2 1 General design of the SHP Since its creation in 1999 the SHP survey Living in Switzerland has covered a wide range of subjects and methods in the area of social sciences The survey is conducted annually from September to February by the M I S Trend Institute in Lausanne and Bern using the computer assisted telephone interview technique CATI The SHP is a panel i e the same persons and households are interviewed year after year and answer with a few exceptions the same questions In contrast to a rotating panel it is an indefinite life simple panel There are therefore no regular activations of new units to be surveyed neither are certain other units withdrawn by choice At present the SHP can see the parallel progress over time of two samples the SHP_ the sample of households and individuals selected in 1999 and interviewed for the first time that year and the SHP II
53. el 1997 1 To ensure a solid database for social reporting on stability and changes in living arrangements and well being in Switzerland that complements data collected by the Swiss Federal Statistical Office 2 To promote opportunities for quantitative social science research by making high quality data available to Swiss social scientists and to the international social sci ence research community The structure of the SHP was formulated using international knowledge of the social sci ences and the experiences of various panel surveys in Europe and North America Bu dowski et al 1998 Budowski et al 2001 Joye and Scherpenzeel 1997 It was based on theoretical work related to the structure and development of contemporary societies Beck 1986 Eisenstadt 1990 Haferkamp 1990 Konietzka 1995 Leisering and Walker 1998 Mayer 1991 Muller and Schmid 1995 analyses of Swiss society and the way it functions H pflinger et al 1991 Leu et al 1997 Levy et al 1997 and on literature about social monitoring Davies 1994 Noll 1998 Like other households panels the SHP is a tool for fine tuning conceptions and analyses of social dynamics Although the dynamics at the macrosocial level are not directly ob served in a panel survey panel surveys are intended to investigate the effects of these changes on the living conditions of households and individuals the manner in which these changes affect the individuals and households and how t
54. employment and quality of life later in time e Emotional trait stability over time How do changes in living conditions and or health affect negative anxiety irritation depressions and positive emotional states joy hope optimism Does a negative emotional state cause illness and low life satisfaction Evidence based answers to these and other questions are highly valuable for the formu lation and implementation of new policies since they facilitate political decision making by means of increased transparency The release of each consecutive wave of SHP data and the synergies between researchers working with the data is leading to a stead ily increasing number of scientific publications All SHP data users are contractually required to report back any publication based on the SHP data be it articles in journals books working papers etc This also applies to unpublished work such as diplomas doctoral theses or seminar work Figure 2 shows the evolution of the number of publications by types since 1999 We highlight the grow ing share of journal articles Figure 1 2 Evolution of publications by types since 1999 100 90 80 70 o Journal Articles a Thesis Master PhD Book Chapters 60 50 40 m Report and Working Paper 30 20 10 0 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 1 4 SHP and CNEF The SHP is currently part of the Cross National Equivalent Fil
55. evel Missing values have been imputed A contract for CNEF variables can be downloaded from www swisspanel ch The documentation on CNEF variables can be found in the SHP Working Paper 5_09 download from www swisspanel ch under publica tions Original responses on the questionnaire are available from the SHP team upon request email to ursina kuhn fors unil ch 5 3 6 Geographical information In addition to the region REGION 7 regions and the canton CANTON 26 can tons in which the household resides two community typologies are constructed This variable is constructed based on the political municipality codes provided by the Swiss Statistical Office see Schuler Dessemontet amp Joye 2005 116f and recoded into 22 codes based on the municipality in which the household is located communes or Ge meinden An aggregated version of this variable in 9 categories is provided as well Ta ble 5 3 13 provides the names and labels of these variables as well as how COM1_ is aggregated into COM2_ 53 Table 5 3 13 Coding of the community typology variables COM1_ COM2_ 1 Great urban centres 1 Centres 1 2 3 2 Median sized urban centres 3 Small centres 4 Centre of peripheral region 5 Wealthy communes 3 Wealthy communes 5 6 Tourist communes 5 Tourist communes 6 7 7 Semi tourist commune 8 Communes with homes and asylums 9 Labour job communes in large central re 2 Suburban communes 9 10 12 13 gi
56. ew 21 489 512 Bendig A W 1954 Transmitted information and the length of rating scales Journal of Experimental Psychology 47 303 308 Bergman M M amp Joye D 2001 Comparing Social Stratifications Schemas CAMSIS CSP CH Goldthorpe ISCO 88 Treiman and Wright Cambridge Studies in Social Research 9 1 37 Bergman M M Lambert P Prandy K amp Joye D 2002 Theorization Construction and Validation of a Social Stratification Scale Cambridge Social Interaction and Stratifi cation Scale CAMSIS for Switzerland Swiss Journal of Sociology 28 7 25 Budowski M Niklowitz M Scherpenzeel A Tillmann R Wernli B amp Zimmermann E 1998 Description of life domains and indicators of the Swiss Household Panel SHP Working Paper 2 98 Neuchatel Swiss Household Panel Budowski M Tillmann R Zimmermann E Wernli B Scherpenzeel A amp Gabadin ho A 2001 The Swiss Household Panel 1999 2003 Data for research on micro social change ZUMA Nachrichten 50 100 125 Budowski M amp Wernli B 2004 Echantillon et taux de r ponse de l exp rience m thodologique 2001 et du questionnaire biographique 2002 SHP Working Paper 2 04 Neuchatel Swiss Household Panel Cauchon C amp Latouche M 2006 Weighting of the Swiss Household Panel SHP Wave 6 SHP II Wave 1 SHP I and SHP II combined Statistique Canada Davies R B 1994 From Cross Sectional to Longitudinal Ana
57. examined to make sure there is information on all household members and the number of household members adds up to the same number as in the household questionnaire Also the variable related to re sponse status is checked Finally demographical variables are checked for consistency with earlier waves This is done for gender date of birth and civil status For other variables the general rule is not to make changes retrospectively i e when in a later wave of data collection an error is found in an earlier wave this is not corrected for the earlier wave 32 CHAPTER 5 DATA DOCUMENTATION 5 1 Data files For every wave every year the data users find a household and an individual file In addition to these annual files there are several other files a household master file an individual master file a calendar file a file containing information on respondents last paid jobs and a social origin file All files are available in SAS STATA and SPSS format 5 1 1 Master files households and individuals The master files of households and of individuals include all households and individual respondents that are in the panel or have been in the panel in the past The files contain an overview of response status for all waves The household master file SHP_MH contains all households of both samples of the panel For every wave it is documented who the reference person is what interviews have been carried out and when they have taken pl
58. fact that a household or an individual is living below a defined poverty line In other words by observing the same individuals over time it is not only possible to study the change in numbers but also the flow of movements between the various states of being and to establish links of causality between different factors and events Moreover the SHP has two other main characteristics that increase its analytic potential 1 it is a comprehensive survey covering a broad range of social fields and a variety of topics allowing for cross domain analysis and 2 all members of the households in the panel aged 14 years and over are inter viewed allowing for intra household studies 1 2 Organisation and funding To date the SHP has experienced three main periods In its first phase 1998 2003 created by the Swiss Priority Program Switzerland Towards the Future the SHP was a joint project run by the Swiss National Science Foundation the Swiss Federal Statistical Office and the University of Neuchatel At the end of the SPP Switzerland Towards the Future the SHP has entered its second phase 2004 2007 Still located at the Univer sity of Neuchatel the SHP has developed a joint venture project Living in Switzerland 2020 aimed at conducting the Statistics of Income and Living Conditions SILC pilot study 2004 2005 in collaboration with the Swiss Federal Statistical Office The SILC pilot data were distributed by the SHP until
59. fer ence points O and 10 These reference points have been given labels that clearly indi cate the end point of the scale for example completely satisfied and not for example very satisfied Scales with two or more reference points and clear labels that fix the end points have proven to decrease the measurement error that can result from variation in response functions Saris and De Rooij 1988 3 Reliability of the data less measurement error Another argument is the effect of measurement error or the reliability of the data Scales with more response alternatives will be more reliable than those with fewer It is often stated that the reliability of scales increases with the number of points used There is probably a limit to the benefit of adding response categories or scale points An interna tional study of satisfaction across 10 different countries showed that the 11 point scale was the most valid and reliable scale of all scales included in the study Scherpenzeel and Saris 1995 In addition the reasons why this type of scale is especially suitable for CATI are 4 Time saving The number production scales do not consist of lists of alternatives that all have to be read aloud in a telephone interview Instead only the first and end point are read aloud and respondents are asked to produce a response alternative themselves This takes considerably less time than reading lists of fully labeled categories 5 No res
60. fferent response patterns A se lection is made of respondents who have participated in an individual interview at least once and who have not left the panel i e not deceased institutionalized or out of the country A distinction is made between respondents who are interviewed in every wave those who are interviewed irregularly and those who dropped out of the panel this implies the respondent is not there in the last three waves SHP I or the last two waves SHP Il Note that calculations are based on unweighted data Table C 1 Demographic characteristics and social involvement attitudes and behaviour by response pattern SHP 1 1999 2008 Always in Ever out lost n 2 630 n 2 856 n 4 675 Sex men 41 10 47 69 47 02 women 58 90 52 31 52 98 Age 14 to 19 17 38 24 86 16 30 20 to 29 10 15 13 06 19 42 30 to 39 23 38 19 61 19 40 40 to 49 18 82 17 75 17 18 50 to 59 15 74 12 96 12 11 60 14 52 11 76 15 59 Education compulsory school 27 72 36 52 31 15 upper secondary level vocational 37 19 35 15 40 35 upper secondary level matura 10 57 9 78 10 27 tertiary level vocational 12 85 10 10 9 58 tertiary level university 11 67 8 44 8 64 Swiss nationality 94 45 91 49 87 10 Region Lake Geneva 16 46 17 82 18 01 Middleland 26 08 23 70 25 30 North west Switzerland 15 67 13 76 14 18 Zurich 18 21 16 18 15 61 East Switzerland 10 46 14 92 13 95 Central Switzerland 9 54 9 35 8 56 Ticino 3 57 4 27 4 39 Urbanization h
61. he number of addresses selected This is also computed at the level of the great regions The pi is the probability computed by a logit model of the household responding to the household questionnaire given response to the grid The p is the probability that at least one member of the household responded to an individual questionnaire given that the household responded to the household questionnaire again modeled by logit Finally f is a factor that is the same for the all households that assures that the weighted sum of individuals is equal to the estimated population total in 1999 according to ESPOP of the Swiss Federal Statistical Office POIDINIT 1 4 2 2b The initial weights for SHP II The SHP II followed all of the households that responded to at least the grid in wave one There was also a non response questionnaire for those who did not respond to the grid This non response questionnaire was used in the modeling of the non response at the grid level As a consequence the initial weight had two levels of adjustment to the design weight The formula for the initial weight is given by 1 POIDINIT m Pi Pi The terms are nearly the same as above where p is the adjustment for responding to 2 the grid or non response questionnaire and p is the adjustment for the responding to the grid given responding to at least the non response questionnaire 4 2 3 Wave specific weights All the weights ar
62. he right to vote d Asked from 2002 onwards 67 Table C 2 Demographic characteristics and social involvement attitudes and behaviour by response pattern SHP II 2004 2008 Always in Ever out lost n 1 500 n 1 315 n 1 569 Sex men 43 3 47 3 47 8 women 56 7 52 7 52 2 Age 14 to 19 13 1 19 5 14 2 20 to 29 7 7 15 7 18 0 30 to 39 19 7 16 9 17 6 40 to 49 22 7 17 9 19 0 50 to 59 15 7 13 6 14 0 60 21 0 16 3 17 2 Education compulsory school 20 6 27 8 26 8 upper secondary level vocational 37 8 38 1 39 8 upper secondary level matura 11 5 9 3 10 1 tertiary level vocational 17 5 14 7 14 0 tertiary level university 12 6 10 1 9 3 Swiss nationality 90 1 78 6 81 0 Region Lake Geneva 17 0 17 3 19 7 Middleland 24 4 25 2 23 4 North west Switzerland 13 3 13 1 14 2 Zurich 20 6 18 3 16 7 East Switzerland 12 5 13 4 13 0 Central Switzerland 9 6 9 3 9 3 Ticino 2 7 3 4 3 8 Urbanization highly and moderately urbanized centres 64 9 62 9 63 1 small urban centres 8 8 10 1 10 0 communes of urbanized centres 10 0 9 7 10 2 communes of small urban centres 7 2 6 0 6 7 communes remote from urbanized centres 9 1 11 3 10 0 Civil status single never married 30 7 42 7 39 8 married 55 1 45 2 46 2 separated 1 7 2 3 1 6 divorced 7 4 6 2 7 5 widower widow 5 1 3 7 4 9 Children in household 51 3 54 6 52 7 Employment active occupied 67 8 67 5 67 7 unemployed 1 4 3 4 3 3 not in labour force 30 8 29 1 29 1 Owner residence
63. hey produce social change on a microsocial level The main purpose of household panels is therefore to understand the processes causes and effects of social changes Of course panel surveys are not the only tools used to measure social change A re peated cross sectional survey makes it possible to calculate for example net transitions between two dates e g a drop in the proportion of the population considered poor or a 1 Panel data are data collected from the same units at more than one point in time It allows for insights into dynamic transformations social processes and changes across time Instead of simply taking a snapshot of people and households at one given point in time by interviewing the same households and their members annually panel data enables the following the observation of changes for the same entities the reconstruction of the nature and development of their ac tions the examination of precedents concurrent dynamics and the consequences of alternative strategies rise in unemployment but not gross transitions e g the number of unemployed still without a job one year later The data collected from household panels supplies unique information allowing not only to estimate gross transitions but also providing an under standing of the transitions observed i e the circumstances family events a change in the activity status heath events etc causing movements in and out of a given state e g the
64. hone number is still valid Contacting mobile phone e mail address or auxiliary Searching directories and the local inhabitant register Request the dcl data care a service of the Swiss post mandated to seek cur rently valid household addresses and the corresponding phone numbers e If no phone number can be found a form is sent to the address provided by the dcl data care asking to complete contact details 3 4 Quality control Prior to each wave extensive pretests are carried out checking correct technical func tioning of filters and new items and running different scenarios After the training of supervisors and interviewers for more details see 3 2 the fieldwork agency monitors the interviewer performance during the fieldwork supervisors listen in to the interviews evaluate interviewers on several criteria e g accurateness and pace of reading argumentation document performance and give feedback to the interview ers M LS Trend carries out the training and monitoring of interviewers in collaboration with the SHP Team 20 4 DATA QUALITY 4 1 Response rates and attrition 4 1 1 Response rates Tables 4 1 and 4 2 indicate the number of validly interviewed households and persons for the years 1999 2008 See Appendix A for further detail on response figures SHP_I With respect to the first sample SHP_I waves 1 to 10 5 074 households were first in terviewed in 1999 In the tenth wave 2 718 households and 4 494
65. i e is transversal or considers multiple years and is longitudinal in nature If the study is done at the household level only transversal weights are provided This is be cause the household is a dynamic unit difficult to define longitudinally Table 1 gives the names of the weights in the SHP and their reference 30 For each of the four types of delivered weights there are two weights produced One is to give the weighted size of the sample the size of the relevant Swiss population These are the weights as described in the constructions above These weights should be used when looking for population totals The second is to maintain the sample size That is to say that the weighted sum of sample members is equal to the un weighted sum These weights should be used when running regressions particularly logistic regressions These weights differ by multiplication of a constant factor only Table 1 gives a list of the names of all the weight variables as they appear in the data sets Furthermore it de scribes their primary use One should note that resident refers to the non institutionalized population residing in Switzerland Table 4 5 List of weights contained in the dataset variable names and description Types of weights Variable name Description Longitudinal weights SHP individuals wp LP1P Weights for longitudinal adults expanded to the resident Swiss population of 1999 wp LP1S Weights expanded to the sample size of longitudi na
66. ighly and moderately urbanized centres 58 75 57 21 59 68 small urban centres 9 24 10 82 10 95 communes of urbanized centres 12 89 10 64 11 17 communes of small urban centres 8 63 8 82 6 78 communes remote from urbanized centres 10 49 12 50 11 42 Civil status 66 single never married 34 60 42 79 39 82 married 54 26 47 37 47 99 separated 1 37 1 33 1 07 divorced 6 46 5 88 6 25 widower widow 3 31 2 63 4 88 Children in household 57 19 64 80 56 40 Employment active occupied 61 37 62 85 64 71 unemployed 1 14 1 68 2 10 not in labour force 37 49 35 47 33 20 Owner residence 51 98 50 14 44 30 Mean satisfaction with health 0 10 8 32 8 25 8 16 sd 1 70 1 83 1 90 Participate in clubs 59 77 52 80 47 18 Mean participated in polls 0 10 7 77 7 14 6 66 sd 2 95 3 17 3 46 Mean general trust in people 0 10 5 91 5 62 5 39 sd 2 35 2 37 2 52 Mean interest in politics 0 10 5 43 5 07 4 71 sd 2 79 2 81 2 94 Mean political influence 0 10 3 48 3 43 3 16 sd 2 60 2 61 2 71 Mean trust in government 0 10 6 00 5 83 5 65 sd 2 03 2 16 2 34 Out of sample left the country institutionalized P Region Lake Geneva VD VS GE Middleland BE FR SO NE JU North west Switzerland BS BL AG Z rich East Switzerland GL SH AR Al SG GR TG Central Switzerland LU UR SZ OW NW ZG Ticino See Appendix A for a list of cantons Only asked to respondents with t
67. in clubs or other groups Voluntary work Association membership Local or parents Association membership Sports or leisure Association membership Culture Association membership Syndicate Association membership Charitable organisation Association membership Women Interest in politics Trust in Federal Government Participation in federal polls Future Active in boycott Future Active in strike Future Active in demonstration Party choice in case of elections tomorrow Talking about politics Frequency Trust in political parties Trust in organisations for the defence of human rights Signing initiative Last 12 months Participating in political meetings Last 12 months Opinion on Swiss tradition direction Seeking job Last four weeks CMJ Part time work Reason years affected by bias 2004 2006 2005 2004 2008 2007 2005 2004 2008 2007 2005 2004 2003 2002 2008 2007 2006 2005 2004 2007 2006 2004 2004 2008 2007 2006 2004 2005 2004 2006 2004 2003 2002 2004 2006 2005 2006 2004 2003 2008 2007 2006 2005 2004 2007 2006 2005 2004 2004 2005 2004 2004 2008 2004 2007 2006 2005 2004 2008 2006 2005 2004 2006 2005 2004 2008 2007 2006 2005 2004 2003 2002 2008 2008 2008 2008 2007 2006 2005 2004 2004 2008 2007 2005 2008 2006 2004 2003 25 4 2 Construction of weights In this section we discuss the weighting procedure for th
68. in person or by proxy in any of the waves since 1999 The information on the last job is collected within the individual interview if the following three conditions are simultaneously met e The person is interviewed for the first time e The person does not currently work P W01 P WO2 and P W03 1 e The person has already worked in a regular way in the past P W07 1 The information on the last job may also be collected in a proxy interview if the following three conditions are simultaneously met e Itis the person s first proxy which is done by an adult e The person does not work i e in the household grid G OCC 1 or 2 e The person has already worked for at least one year X W05 Because this information is collected only once it is not necessary to display it in the in dividual file in every wave The information is rather combined in a file last job com prising the variables of the individual questionnaire and the proxy questionnaire in which the wave identifier is renamed by A separate variable LUYY indicates the wave in which the information is collected Note that if a respondent is not working at a given wave but has been working in any of the previous waves this information is not included in the last job file but has to be taken from the previous annual individual files 5 1 5 Social origin file This file contains information on the employment status of the parents when the respon dent was
69. includes households whose numbers are not listed or households that could not be contacted by telephone This problem may produce a bias namely differences between the estimates based on the observed population SHP survey and those that would have been produced based on the target population all individuals living in private households in Switzerland at a given time e duplicates despite meticulous checking of the SRH to ensure that only one num ber is kept per household some households appear more than once in the sur vey frame This problem results in wrong initial selection probabilities In spite of this a correction factor is not calculated for households with several telephone lines The information is available but the effect is negligible e overcoverage selection of units outside the target population businesses homes prisons collective households second homes It should be noted that for a panel this problem is only encountered at wave 1 and that these ad dresses are usually considered as out of sample non sample cases 2 3 Following rules 2 3 1 Initial rules governing contact with households The general rule is to interview all households that completed at least the grid during the previous wave We proceed with interviews as long as members of these households agree to fill in the household questionnaire and his or her individual questionnaire it is always possible to catch up with the other individuals in
70. istakes high income increases or decreases with respect to the last wave extreme income inconsistencies between the sum of income sources and total income and inconsistencies between individual and household income see www swisspanel ch for details Individual income Table 5 3 10 List of constructed income variables of individuals Variable Gross net Description I EMPY EMPYG gross Income from employment annual amount I EMPYN net social Takes account of 13 and 14 month salary bonuses and grati security contributions de fications see www swisspanel ch ducted I INDY I INDYG gross Income from self employment annual amount I INDYN net social se Takes account of 13 and 14 month salary bonuses and grati curity contributions de fications if applicable see www swisspanel ch ducted ISSOASIY State pension for old age first pillar widow er s or orphans annual amount Includes additional benefits S AIY Disability pension annual amount Includes additional benefits I PENY Income from pension schemes second pillar old age pension annual amount Includes additional benefits I UNEY Income from unemployment social insurance annual amount I WELY Income from welfare benefits social assistance annual amount I GRAY Income from scholarships grants annual amount Income from private or public institution ISSINSY Income from any another private or public institution annual amount ISSFAMY Family or child allow
71. ividual questionnaire or if the sum of individual income is larger than the household income from the household questionnaire In the other cases 52 household income from the household interview is taken Only if household income is based on individual income adjustments are made for gross and net income Table 5 3 12 List of constructed income variables of households Variable Gross net Description ISSHTY I HTYG gross Yearly income from all members I HTYN net Taxes not deducted social security taken account of where possible I EQS I EQSG gross Yearly household income equivalised ac I EQON net cording to SKOS scale 1998 see social security taken account of www swisspanel ch where possible Taxes not deducted I EQO I EQOG gross Yearly household income equivalised ac I EQON net cording to modified OECD scale see social security taken account of www swisspanel ch where possible Taxes not deducted Additional Income variables The constructed annualised income variables of the SHP user files have been imputed if the amount was missing don t know no answer implausible value These imputed values are available from the SHP team upon request email to ursina kuhn fors unil ch The SHP cross national equivalent file CNEF contains income sources defined slightly differently than in the SHP user file The CNEF variables with the ex ception of professional income report income on the household l
72. l Sciences 07 025 Beverly Hills Sage Mayer K U 1991 Soziale Ungleichheit und die Differenzierung von Lebensverlaufen in W Zapf Ed Die Modernisierung moderner Gesellschaften Verhandlung des 25 Deutschen Soziologentages 1990 pp 667 687 Frankfurt am Main Westdeutscher Verlag M ller H P amp Schmid M 1995 Sozialer Wandel Modellbildung und theoretische Ans tze Frankfurt am Main Suhrkamp Noll H H 1998 Die Perspektive der Sozialberichterstattung in P Flora amp H H Noll Eds Sozialberichterstattung und Sozialstaatbeobachtung pp 13 28 Frankfurt Cam pus Verlag Paugam S 2000 Le salari de la pr carit Les nouvelles formes de l int gration pro fessionnelle Paris Presses Universitaires de France 60 Plaza S and Graf E 2007 Recommandations et exemples pratiques concernant l application des pond rations Neuchatel Swiss Household Panel Saris W E and De Rooij K 1988 What kind of terms should be used for reference points In Saris W E Ed Variation in response functions A source of measurement error in attitude research Amsterdam Sociometric Research Foundation Scherpenzeel A C and Saris W E 1995 The quality of indicators of satisfaction across Europe A meta analysis of multitrait multimethod studies In A C Scherpenzeel Ed A question of quality Evaluating survey questions by multitrait multimethod studies Dissertation Leidschendam Royal PTT Ne
73. l adults in the first panel SHP I and SHP II combined wp L1P Weights for longitudinal adults expanded to the individuals resident Swiss population of 2004 wp L1S Weights expanded to the sample size of longitudi nal adults in the combined panels Transversal weights SHP and SHP II combined wp T1P Weights expanded to the resident Swiss popula individuals tion of current year wp T1S Weights expanded to the sample size of the com bined panels SHP I and SHP II combined wh T1P Weights expanded to the resident Swiss popula households tion of current year wh T1S Weights expanded to the sample size of individu als in the households Note corresponds to the two last digits of the year in question One should note that the longitudinal weights make reference to the first year that is 1999 for the first panel and 2004 for the combined panel However it is generally better to use a slightly imperfect longitudinal weight which will at least take into account inclu sion probabilities and non response then none at all 4 2 6 Complex sample Weighting provides estimates that are representative of the national population Another issue has to be considered when using the SHP the complex sample structure of the data The standard procedures of common statistical software packages e g SAS SPSS STATA underestimate variance Plaza and Graf 2007 because they assume a simple random sample As most with most surveys the SHP sample
74. lar for both panels after five waves SHP_1 2003 and SHP_II 2008 Table 4 1 Number of households validly interviewed in SHP_I and SHP_II 1999 2008 Year Wave SHP_In SHP_II SHP_I ll A B n A B n 1999 1 5 074 100 100 2000 2 4 425 87 87 2001 3 4 139 82 94 2002 4 3 582 71 87 2003 5 3 227 64 90 2004 6 1 2 837 56 88 2 538 100 100 5 375 2005 7 2 2 457 48 87 1 799 71 71 4 256 2006 8 3 2 537 50 103 1 684 66 94 4 221 2007 9 4 2 817 56 111 1 494 58 89 4 311 2008 10 5 2 718 54 96 1 546 61 103 4 264 Table 4 2 Number of persons validly interviewed in SHP_I and SHP_II 1999 2008 Year Wave SHP I SHP_Il SHP_I ll SHP_I fully n A B n A B n longitudinal A B n 1999 1 7 799 100 100 7 799 100 100 2000 2 7 073 91 91 6 335 81 81 2001 3 6 601 85 93 5 429 69 86 2002 4 5 700 73 86 4 480 57 83 2003 5 5 220 67 92 3 888 50 87 2004 6 1 4 413 57 85 3 654 100 100 8 067 3 076 39 79 2005 7 2 3 888 50 88 2 649 72 72 6 537 2 622 34 85 2006 8 3 4091 52 105 2 568 70 97 6 659 2 399 31 91 2007 9 4 4 630 59 113 2 350 64 92 6 980 2 209 28 92 2008 10 5 4 494 58 97 2 410 66 103 6 904 2 060 26 93 These percentages are calculated on the basis of the number of interviews conducted in the first year 1999 or 2004 These percentages are calculated on the basis of the number of interviews conducted in the previous year They may therefore exceed 100 Since 2006 the number of interviews increases due to various measures t
75. le Label Constructed from HHMOVE moved since last interview grid and M I S Trend information 54 5 4 Missing value conventions The following missing value labels are used 1 does not know 2 no answer 3 inapplicable This means either a the specific question was not applicable for the respondent b the respondent did not participate in this particular wave c the entire household did not respond was not contacted 7 filter error a question should have been asked but was not 8 other error 5 5 Imputation procedures Apart from the consistency checks and corrections see 4 4 no values are changed or imputed Upon request imputed values on most income variables are available contact ursina kuhn fors unil ch 5 6 Combining data files Table 5 6 1 shows the identification numbers that are available in the different data files The personal ID idpers can be found in all files on the individual level always referring to the same individual The interviewer ID is available in the interviewer files see 5 1 7 and the annual individual and household files As the composition of households can change over time their identification number is wave specific Identification numbers of parents and spouses refer to their personal ID For example to match parents and children one can attach the info of the parent to the info of the child by matching idmoth and idfath idmoth__ and idfath_ in STATA and SAS to id pers To co
76. lysis In A Dale amp R B Davies Eds Analyzing Social amp Political Change pp 20 40 London SAGE Eisenstadt S 1990 Kultur und Sozialstruktur in der neueren soziologischen Analyse In H Haferkamp Ed Sozialstruktur und Kultur pp 7 20 Frankfurt am Main Suhr kamp Eurostat 2003 ECHP UDB description of variables Data Dictionnary codebook and differences between countries and waves Eurostat http circa europa eu Public irc dsis echpanel library l user_db pan166200312pdf _ EN 1 0 amp a d 58 Farago P 1996 Gesellschaftliche dauerbeobachtung im SP Zukunft Schweiz Demain la Suisse Swiss Priority Programme Switzerland towards the future Bern Swiss Na tional Science Foundation Frick J R Jenkins S P Lillard D R Lipps O amp Wooden M 2007 The Cross National Equivalent File CNEF and its Member Country Household Panel Studies Schmollers Jahrbuch 127 627 654 Ganzeboom H B G amp Treiman D J 2003 Three Internationally Standardised Measures for Comparative Research on Occupational Status In J H P Hoffmeyer Zlotnik amp C Wolf Eds Advances in cross national comparison A European working book for demographic and socio economic variables pp 159 193 New York Kluwer Academic Press Garner W R 1960 Rating scales Discriminability and information transmission Psychological Review 67 343 352 Goldthorpe J H 1987 Social Mobility and
77. m unemployed Unemployed The primary distinction in an employment relations approach is that between employers who buy the labour of others and assume some degree of authority and control over them self employed or own account workers who neither buy labour nor sell their la bour to others and employees who sell their labour to employers Employees are further differentiated according to the employment relations of their oc cupation employers are separated by size of establishment and the self employed ac cording to occupation Broadly speaking the kind of contracts employees have depend upon a how easily their work may be monitored and controlled by the employer and b asset specificity i e how specific and crucial their knowledge of technical and organiza tional issues is to the employer When monitoring is difficult and asset specificity is high a service relationship will be typical labour contracts apply where labour is more easily replaceable in these terms A complete user guide of the ESeC can be downloaded here http www iser essex ac uk research esec user gquide 48 D The Swiss Socio Professional Categories CSP CH The Swiss Socio Professional Categories CSP CH Joye amp Schuler 1995 are based on the occupational coding of the Swiss Federal Office of Statistics as well as educa tional achievement The logic of the CSP CH is as follows Table 5 3 9 Swiss Socio Professional Categories Universit
78. mbine information from the household reference person with the household refper needs to be matched to idpers in the individual file To add information from the partner to this file rpspou needs to be matched to idpers 55 Table 5 6 1 Identification numbers variable In files description idint P H V ID of interviewer Idpers P MP SO CA LJ BH ID of person BV Idhous P H MP MH BH ID of household Idfath MP ID of father Idmoth MP ID of mother Idspou P ID of partner Refper H MH ID of reference person in hld Rpspou H ID of partner of reference person a P individual questionnaire wave specific H household questionnaire wave specific MP master file individuals MH master file households V interviewer file SO social origin CA activity calendar LJ last job BH biographical file horizontal BV biographical file vertical P Attention The values of the variable idint in the Interviewer data files have been coded in order to protect the identity of the Interviewers Consequently the merging of the Interviewer data with the Household and Individual level files is only possible after de coding Please contact Oliver Lipps for more details oliver lipps fors unil ch On www swisspanel ch there are examples of programming in SAS SPSS and STATA of how to combine different files such as matching respondents across waves matching respondents to households matching couples etc 5 7 Changing the l
79. mployees are introduced instead Because the calendar file contains information from all waves some detail present in the separate waves has been lost The calendar file does not include a distinction between small and large part time jobs nor does it have a distinction between self employed indi viduals and employees Users of the data interested in analyzing these distinctions are advised to use the calendar questions in the personal files of the appropriate waves In the calendar file the following codes are applied 1 Employed full time 2 Employed part time 3 Unemployed 4 Inactive 5 Unemployed or inactive relevant for inactive respondents in W2 and W3 only Table 5 1 1 shows the different versions of the calendar questions in the individual inter views and the corresponding codes in the calendar file 35 Table 5 1 1 Questions in the personal questionnaire related to the activity calendar and the corresponding codes in the calendar file W2 and W3 W4 and W5 W6 to present Original question Cal Original question Cal Original question Cal Original question Calen Employed respondents endar Inactive respondents endar endar dar value value value value We are going to review the We are going to review the We are going to review the We are going to review the months months between now and months between now and months between now and since month year and for each month year and for each month year and for each month year
80. ng frame 2 3 Following rules 2 3 1 Initial rules governing contact with households 2 3 2 Initial rules governing the follow up of individuals 2 3 3 Additional rules to fight attrition 2 4 Questionnaires 2 4 1 Content of the questionnaires 2 4 2 The use of 11 point scales Fieldwork 3 1 Approaching the participating households 3 2 Selection and training of interviewers and supervisors 3 3 Measures to increase response 3 3 1 Incentives for the interviewers 3 3 2 Incentives for the participating households 3 3 3 Refusal conversion 3 3 4 Maintaining contact with respondents 3 3 5 Minimizing noncontact 3 4 Quality control Data quality 4 1 Response rates and attrition 4 1 1 Response rates 4 1 2 Attrition 4 2 Construction of weights 4 2 1 Overview of techniques 4 2 1a The process of segmentation 4 2 1b The calibrations 4 2 2 Formation of initial weights or baseline weights 4 2 2a The initial weights for SHP 4 2 2b The initial weights for SHP II OMNDAHRWO 4 2 3 Wave specific weights 4 2 3a Individual longitudinal weights 4 2 3b Individual transversal weights 4 2 3c Household transversal weights 4 2 4 4 2 5 4 2 6 Weights for the combination of the two panels Selection of the appropriate weight Complex sample 4 3 Data cleaning Consistency checks and corrections 5 Data documentation 5 1 Data files Master files households and individuals Annual files households and individuals Calen
81. ny status Other person Therefore valid values are only available for the persons interviewed for the first time in wave 1 For all the others theses values are labelled missing The questions regarding the parents political orientation are asked since wave 4 2002 P P46 Political position Left Right Father P P47 Political position Left Right Mother In wave 4 every person responding to the individual questionnaire was asked these two questions in order to obtain this information also from persons having already been in terviewed in previous waves in which the questions weren t posed Since wave 5 these two questions are part of the social origin module and are addressed only to persons who are interviewed for the first time Consequently the information is missing for per sons who answered the social origin module before wave 4 and who did not participate in wave 4 5 1 6 Biographical files To obtain at a single occasion additional information about the respondents life course prior to the panel study a retrospective biographical questionnaire was developed with questions regarding educational working and family history SHP Questionnaires Biography This paper and pencil questionnaire was sent to the respondents by mail and was self administered Biographical information was gathered in the following domains Living arrangements LA Periods outside of Switzerland SA Changes in civil status CS
82. o participate at the survey They work through the questionnaires and study the training manual as well as the advance letters and newsletters which the participating households received The training sessions are conducted by M I S Trend in Lausanne and Bern with the as sistance of the supervisors and a member of the SHP Team For the refusal conversion only the most successful interviewers measured by their individual response rate and the quality of their interviewing performance are used Extra training is conducted to be well prepared 18 M LS Trend ensures a strict selection of only the most experienced interviewers and guarantees that all interviews are conducted by native speakers 3 3 Measures to increase response Over the past years the SHP has taken several measures to fight attrition These meas ures concern incentives for the interviewers incentives for the participating households refusal conversion maintaining contact with the households and minimizing noncontact 3 3 1 Incentives for the interviewers To increase the interviewers motivation they can earn two collective bonuses One bo nus is based on the general response rate interviewers have to accomplish at least 95 of last year s individual interviews The second bonus is oriented towards interviewers who are engaged in refusal calls only and is based on the refusal conversion rate Addi tionally there are regular briefings of all interviewer
83. ociale en Suisse cat gories 59 socio professionnelle Bern Office f d ral de la statistique Konietzka D 1995 Lebensstile im sozialstrukturellen Kontext Ein theoretischer und empirischer Beitrag zu Analyse soziokultureller Ungleichheiten Opladen Westdeutscher Verlag Kuhn U 2009 Attrition analysis of income data SHP Working Paper 2 09 Lausanne Swiss Household Panel Leisering L amp Walker R 1998 The dnamics of modern society Briston The Policy Press Leu R E Burri S amp Priester T 1997 Lebensqualitat und Armut in der Schweiz Bern Haupt Levy R 2002 Meso social structures and stratification analysis A missing link Revue suisse de sociologie 28 193 215 Levy R Joye D Guye O amp Kaufmann V 1997 Tous gaux De la stratification aux repr sentations Zurich Seismo Lipps O 2006a Attrition and motivation in the Swiss Household Panel SHP Working Paper 2 06 Neuchatel Swiss Household Panel Lipps O 2006b Analysis of panel participation in couples using interviewer character istics and the partner s behaviour SHP Working Paper 3 06 Neuch tel Swiss House hold Panel Lipps O 2006c Attrition in the Swiss Household Panel wave 2 through wave 7 SHP Working Paper 4 06 Neuchatel Swiss Household Panel Lodge M 1981 Magnitude scaling Quantitative measurement of opinions Sage University Paper series on Quantitative Application in the Socia
84. ons 10 Suburban residential communes in large central regions 11 Peripheral urban communes in large central 4 Peripheral urban communes 11 14 regions 12 Labour job communes outside large central regions 13 Suburban residential communes outside large central 14 Peripheral urban communes outside large central regions 15 Net immigration communes moderate or 7 Rural commuter communes 15 16 high proportion 16 Native resident communes moderate or high proportion 17 Communes with industrial and tertiary sec 6 Industrial and tertiary sector communes tor employment 4 8 17 18 18 Communes with industrial employment 19 Communes with agricultural and industrial 8 Mixed agricultural communes 19 20 employment 20 Communes with agricultural and tertiary sector employment 21 Communes with agricultural employment 9 Peripheral agricultural communes 21 22 population 22 Communes with strongly shrinking popula tion The municipality codes themselves are not included to guarantee the anonymity of the respondents Under certain conditions are the codes available for users of the data This requires special authorization and is only possible when anonymity of the households can be guaranteed Other constructed variables in the household file related to socio geographical charac teristics of the household are HHMOVE whether the household moved since the last interview Table 5 3 14 Household moved since last interview HHMOVE Variab
85. ousehold Relationship to other persons in house hold civil status number of persons and children in household Table 5 3 2 Household composition variables in household file Variable name Description MAXCOH Maximum duration of existence of NBADUL NBKID AOLDKI AYOUKI ADUK1_ ADUK2_ NBB_ household in years Number of adults in hid gt 18 Number of children in hid 0 17 Age of oldest coresident child max 17 Age of youngest coresident child max 17 Number of adult children in hid gt 18 amp lt 30 Number of adult children in hid gt 30 New born baby birth between two con secutive grid interviews or within last 12 months if no previous year grid inter view Information used for construction Longest time of two members living together in years information from grid Information from grid Information from grid Information from grid Information from grid Information from grid and individual questionnaire Information from grid and individual questionnaire Information from household and indi vidual master file 42 Table 5 3 3 Socio demographic variables in individual files Variable Description Information used for construction name SS AGE Age in year of interview Collected once confirmed next waves Difference from the year of birth and the official year of interview official means the year ofthe beginning of the wave in question e
86. persons responded Out of the 7 799 persons interviewed for the first time in 1999 26 n 2 060 responded to their personal interview in each of the following waves including the tenth wave con ducted in 2008 At the household level see Table 4 1 A the drop in participation was particularly significant in the second 13 and the fourth 11 waves compared to the other waves in the 1999 2005 period 5 to 8 From 2006 onward the number of households validly interviewed increases in general but note the slight 2008 reduction in the number of interviews conducted due to various measures taken to convert house holds who where abandoned earlier because of double refusals into respondents see for more information the Swiss Household Panel Scientific Report 2008 downloadable from www swisspanel ch under Project Evaluation and Scientific Report At the individual level see Table 4 2 A the drop in participation was particularly sig nificant in the fourth 12 wave as compared to the other waves in the 1999 2005 pe riod between 6 and 10 From 2006 onward the number of persons validly inter viewed increases in general but note the slight 2008 reduction in the number of inter views conducted due to 1 various measures taken to convert households who where abandoned earlier because of double refusals into respondents and 2 efforts made by the interviewers of M I S Trend to enroll all eligible household members for an indivi
87. pisodes of respondents 5 1 7 Interviewer files These files contain information gathered from the interviewers who conducted the SHP interviews by means of paper and pencil questionnaires In all waves except Wave 1 3 and 4 the interviewers answered a short questionnaire The questionnaires measure a number of interviewer characteristics demographic traits such as sex age language and education but also characteristics such as the attitude of the interviewers towards this type of study and towards sensitive questions According to the SHP research inter ests the questionnaires have been changing over time 39 Attention The values of the variable idint in the Interviewer data files have been coded in order to protect the identity of the Interviewers Consequently the merging of the Interviewer data with the Household and Individual level files is only possible after de coding Please contact Oliver Lipps for more details oliver lipps fors unil ch Note further that in 2008 Wave 9 the interviewer ID changed Because three digits to identify interviewers was not enough all interviewers located in the Lausanne studio were added a value of 10 000 and all interviewers located in the Bern studio were added a value of 50 000 This is important for longitudinal interviewer analyses 5 2 Variable naming conventions The variable names are coherent over time The only change is found in the year indica tor In order to assure consistenc
88. ponse order biases Response alternatives presented at the beginning and end of a list may be more likely to be recalled and therefore perhaps selected more often When no visual aids are pre sented and when the list is long memory effects may be important Schuman and Presser 1981 The number production scales do not consist of lists of alternatives Instead only the first and end point are read aloud and respondents are asked to produce a response al ternative themselves Since CATI is exclusively oral verbal category scales are likely to suffer from the response order biases Therefore number production scales are more appropriate in CATI 17 Chapter 3 Fieldwork This chapter provides information on how the fieldwork for the SHP is carried out Start ing with the selection and training of the interviewers we describe how the participating households are approached through to the measures taken to increase response and quality control Since the beginning in 1999 the fieldwork for the Swiss Household Panel SHP is done by M I S Trend in Lausanne and Bern www mistrend ch conducting computer assisted telephone interviews CATI in German French and Italian 3 1 Approaching the participating households The fieldwork is scheduled from September to February and starts with sending a letter to the participating households informing them about the upcoming interviews To make sure that the first personal contact by an interviewer
89. pulation charac teristics discussed above for the baseline year 1999 for the SHP I and 2004 for the SHP Il 4 2 3b Individual transversal weights These weights are for performing cross sectional studies and for obtaining population totals representative for the year in question A weight sharing is performed in house holds that have non original sample members non OSMs The weight share depends on whether the non OSMs were present at baseline 1999 or 2004 If they were pre sent the weight is the same for all individuals of the household and is equal to D P_NRGRIL PTI _ PAR TR 4 where L is the number of longitudinal individuals and P is the number of non OSMs ini tially present If the non OSMs were not present at baseline the weights are P_NRGRIL for longitudinal person j L PTI _ PAR 4 P_NRGRIL 5 fornon OSM initially absent j L Once the weight sharing is done it is adjusted for non response to the individual ques tionnaire using segmentation to form PTI_NRQI The segmentation analysis is per formed using all individuals that are aged 14 in the current wave and in households having responded to the grid and containing at least one longitudinal individual Finally the weight is calibrated to the current year s population estimates using the calibration technique already described 29 4 2 3c Household transversal weights These are designed to perform studies at the household level
90. re the sharing of tasks and decision making within the household The individual questionnaires cover the following topics 1 the household and the family comprising objective elements such as the existence of children living outside the household the sharing of housework and childcare as well as Subjective elements such as satisfaction with private life and with the sharing of the housework 2 health and victimisation covering objective elements such as general illness and health problems visits to the doctor and hospitalisation long term handicaps threats or attacks endured together with subjective elements such as the self perceived state of health the estimated evolution of the state of health or satisfaction with one s own health 3 social origins asked at first interview only referring to information related to profes sion professional position educational level political positioning and the nationality of both parents together with possible financial difficulties in the family of origin 4 education covering the various levels of achieved education education currently be ing pursued fluency in foreign languages and participation in on the job training 5 employment considering four different aspects firstly the collection of information necessary to determine the status of the interviewee in the labour market secondly in formation covering the current main employment thirdly
91. re activities and the use of the media as well as subjective elements such as satisfaction with leisure and free time From the second wave on the questionnaire also includes a life events module and an occupational calendar module covering the 12 months prior to the interview More information on the content of the questionnaires is available here http www swisspanel ch codebook cblqre php lang en amp pid 207 And here as pdf http www swisspanel ch doc q_pdf php lang en amp pid 20 2 4 2 The use of 11 point scales For many questions of the Swiss Household Panel questionnaire the 11point scale has been chosen instead of a category scale The 11point scale is used in many other ongo ing surveys for example the GSOEP and World Value Study and seems to be well handled by respondents Respondents are asked to indicate the strength of their attitude or opinion in a number between 0 and 10 with the endpoints 0 and 10 being defined by verbal labels This type of scale is often called a number production scale The main arguments in favour of this type of scale are 1 Minimisation of categorisation effects We assume that attitudes fall along a single latent continuum ranging from positive to negative The larger the number of points on a response scale the better it represents this underlying latent continuum and the more accurate it reflects the variation Scales with relatively few response alternatives for
92. repre sents the combination of occupations among couples who have the same occupational title typical examples are couples who both work together on a farm or a restaurant The second dimension represents the social distance that is reflected in the typicality of pairings between couples occupations It should be added that the scores of a dimen sional analysis do not have sociological significance in themselves but only in relation to each other Here the value allotted to each occupation i e the score of the dimensional analysis indicates its position on this hypothetical social axis and consequently its dis tance to others Subsequently each occupation of the 4 digit ISCO 88 classification is allotted a CAMSIS score The current version adjusts for national variations and is sensi tive to gender Other dimensions can be easily accommodated e g ethnicity geo graphic region in order to incorporate specific research interests and hypotheses and to improve the correspondence between this measure and the social categories within their context See for more information Bergman et al 2002 and Bergman and Joye 2001 5 3 4 Professional integration Paugam s typology is based on a distinction between conditions of employment and conditions of work The typology distinguishes four types of professional integration see Paugam 2000 Secure integration int gration assur e is defined as the combination of job stability and
93. s and supervisors on the progression of the fieldwork 3 3 2 Incentives for the participating households To encourage respondents to participate in the survey one of the following incentives are offered a voucher for the cinema a voucher for the Swiss railways or a donation to charity e g Amnesty International WWF Caritas The incentive is sent to the respondent one week after completing the individual inter view at the latest aiming to encourage other members of the household to also take part in the survey 3 3 3 Refusal conversion Households that have not participated in the survey for one year or more have been re approached progressively These households are sent a preliminary letter with the re quest to take part in the next wave of data collection including a railway voucher Only the most successful and specially trained interviewers are allowed to contact these households Similarly households and individuals who refuse participation in the current wave are re contacted at a later point by refusal conversion trained interviewers The refusal conversion rate amounts to nearly 40 3 3 4 Maintaining contact with respondents To avoid household drop out of the panel because of unsuccessful tracing due to mov ing changed phone numbers households splits etc several measures are taken to ensure that contact can be established with the respondents in new waves First the participating households are informed annuall
94. s in 2000 They are classified by their respective domains in the codebook and are found in the module to which they belong see 5 3 5 3 Constructed variables This paragraph presents background information on the construction of socio demographic variables education labour market participation and income socio geographical information and weights For all other constructed variables we refer to www swisspanel ch under Documentation Variables 41 5 3 1 Socio demographic variables Tables 5 3 1 to 5 3 3 present the constructed socio demographic variables in the house hold file Table 5 3 1 and 5 3 2 and the individual file Table 5 3 3 Table 5 3 1 Household typology variables in household file Variable name HLDTYP HLDFFS HLDCEN Description Type of household Classification adopted from European Community Household Panel Eurostat 2003 and PACO Household typology adopted from the Fertility and Family Survey FFS The FFS was launched by the United Nations Economic Commission for Europe and was commissioned by the Swiss Federal Statistical Office for Switzerland www bfs admin ch Household typology Swiss Census Swiss Federal Statistical Office www bfs admin ch Information used for construction Relationship to other persons in house hold civil status number of persons and children in household Relationship to other persons in house hold civil status number of persons and children in h
95. s with the number for 2008 appearing in parentheses As can be seen by the tables the weights do seem to correct for much of the bias due to attrition particularly for a given year Table 4 3 The cumulative results for all waves combined and 2008 independently Significant Significant Signification and biased variables Number of occur difference difference rences in all of the without with 308 variables tested weight weight 2008 in parenthe ses No No No significant difference with or without 211 231 weights The variable considered does not ap 68 5 75 0 pear to be biased from attrition No Yes No significant difference without the weights but 6 1 the weighted results have a difference The 1 9 0 3 weighting biases the variable Yes No We observe a significant difference without 59 63 weights but it disappears when the results are 19 2 20 5 weighted The variable is therefore touched by attrition but the weighting corrects the phenom ena Yes Yes We observe a significant difference without the 32 13 weight and it persists even with weighting The 10 4 4 2 variable is therefore touched by attrition without the possibility of correction by weighting Mainly leisure and politics variables Table 4 4 lists the 38 variables 6 32 from Table 1 that showed bias after weighting for any year and the years in which they demonstrated bias It should be noted that the ma jority of these variables demonstrated
96. se of the household weight The weights used are those after the final adjustment for non response PL_NRQI PTI_NRQI and PTM_NRQM respectively as described above with one exception The exception is the longitudinal weights for the SHP I Since the non response is computed in reference to 1999 a new correction for non response is calculated making reference to whether the individual was eligible for the selection in the second panel i e was living in Switzerland and non institutionalized in 2004 If the unit is a member of SHP the weight is then multiplied by the factor p If the unit is a member of the SHP II the weight is multiplied by the factor 1 p This means that each sample is multiplied by the ratio of units in the sample After this factor of adjustment is enacted the panels are com bined and then the weights are calibrated for the longitudinal transversal individual and transversal household weights as described above 4 2 5 Selection of the appropriate weight The weights should be used to get results that are representative of the targeted popula tion which is the non institutionalized Swiss population This is particularly true for de scriptive statistics but they are also important to use in regression models when possi ble to obtain coefficients the estimates of error and significance that are nationally rep resentative In the selection of a weight one needs to know whether the study concerns only one year
97. selection is more complex as it has stratification clustering and adjustments due to non response Such complex sample needs to be taken into account in the analysis to obtain appropriate es timates of the variance For SAS users the recommendation is to rely on the survey 31 procedures for example PROC SURVEYFREQ PROC SURVEYMEANS PROC SUR VEYREG PROC SURVEYLOGISTIC For STATA users the commands svyset and svy have to be used For SPSS users the module complex sample is needed 4 3 Data cleaning Consistency checks and corrections Before the data is released a few consistency checks are performed First the filters used in the questionnaire are checked In the rare occasions in which a filter was applied wrongfully a question was either asked when it should not have been or was not asked when it should have been In the first situation the answer to the question is deleted and the value is set to 3 not applicable see missing value conventions In the second situation a code of 7 is given filter error see missing value conventions Second the value range of all questions with restricted response categories is verified Values out of range are usually related to recoding mistakes and are corrected The value ranges of open questions are not scrutinized because setting a limit beyond which point values become highly unlikely is always arbitrary Third the households and their individual members are
98. to respond to the survey old age handicap etc b Referring to all gross households minus those with neutral problems neutral problems invalid telephone etc c Referring to all called individuals minus those with neutral problems foreign language etc Note SHP_I denotes the original households recruited in 1999 63 Table B 2 Participation in the Living in Switzerland Panel Survey 2004 2008 SHP_1 11 Number of partici SHP_I SHP_II SHP_I SHP_II SHP_I SHP_II SHP _I pating units 2004 w6 2004 w1 2005 w7 2005 w2 2006 w8 2006 w3 2007 w9 Households with 2 918 2 704 2 526 1 908 2 580 1 754 2 893 grids completed Household inter 2 837 2 538 2 457 1 799 2 537 1 684 2 817 view completed Persons living in 7 517 6 569 6 491 4 673 6 587 4 276 7 225 participating house holds Persons aged 14 5 976 5 376 5 220 3 845 5 333 3 500 5 972 years and older eli gible for individual interviewing Personal interview 4 413 3 654 3 888 2 649 4 091 2 568 4 630 completed Proxy Interviews a 1 482 1 117 1 241 772 1 237 745 1 226 Persons responding 3 076 2 622 2 395 2 399 1 930 2 209 in current and all previous waves Grid level net re 82 65 91 81 87 78 86 sponse rates b Individual level net 85 76 87 75 81 78 81 response rates c Source Swiss Household Panel 1999 2008 http www swisspanel ch SHP II 2007 w4 1 548 1 494 3 777 3 123 2 350 639 1 601 84 80 SHP I 2008
99. ulation groups who have recently migrated to Switzerland are represented In order to compensate for the contin ual erosion of the original 1999 sample deaths hospitalisation migration refusals a refreshment random sample of new households was injected in 2004 SHP_II follow ing the same methodology 2 2 1 Sampling plan The samples of SHP_I and SHP_II are stratified by major geographic region NUTS II in proportion to the number of households per stratum The addresses of the gross sample are distributed in the following proportions 1990 and 2000 censuses This part is taken from Graf 2009a 10 12 11 Table 2 1 Stratification of gross sample Strata Cantons Proportion of ad Proportion of ad dresses SHP_I dresses SHP_II Lake Geneva region VD VS GE 18 45 18 22 Mittelland BE FR SO NE JU 23 25 22 92 North west Switzer BS BL AG 13 44 13 86 land Zurich ZH 17 51 18 22 Eastern Switzerland GL SH AR Al SG 15 68 13 70 GR TG Central Switzerland LU UR SZ OW NW 7 20 8 75 ZG Ticino TI 4 47 4 33 Total 100 100 See Appendix A for a list of cantons and their abbreviations The selection is strictly proportional to the number of households per major region and does not take into account the average number of persons in the households per region Ticino is not overrepresented The survey was based on the SRH which is updated quarterly in March June Septem ber and December The SHP_ w
100. ven when interview took place beginning of following calendar year CIVSTA Civil status in year of interview Information from household grid and personal interview Equivalent to ques tion P D13 Individual information is considered more reliable than from reference person MAXCOP Max time in years of person living with Information from grid someone else in household NAT_1_ First nationality Grid and individual questionnaire NAT_2_ Second nationality Grid and individual questionnaire NAT_3_ Third nationality Grid and individual questionnaire HAB_CH Duration of residence in CH since Grid and individual questionnaire when G YCH P D164 Please not that gender of the respondent is included in the individual master file see 5 1 1 5 3 2 Education Table 5 3 4 shows the constructed variables related to level of education This list does not include the original or recoded variables related to education For all available vari ables on education we advice to go to our website www swisspanel ch under Documen tation Search by domains select education Table 5 3 4 Constructed variables related to education in the individual files Variable Description Information used for construction name a EDUCAT Highest level of education achieved in From household grid and individual inter 11 codes view Individual interview considered more reliable EDCAT Highest level of education achieved in From household grid and indi
101. vidual inter 17 categories view Individual interview considered more reliable 5 3 3 Work status occupation and social position Work status WSTAT is constructed from P WO1 working for pay last week P WO03 have a job although not working last week and P W06 can start work im mediately from the individual questionnaire 43 All social stratification measures presented below are fundamentally based on the re spondents occupational titles which were coded in great detail by the Swiss Federal Office of Statistics This Swiss specific code was then recoded into the International Standard of Classification of Occupations ISCO 88 developed by the International La bour Office The use of stratification schemas based on occupational titles traditional in this field has as a consequence that only people who report an occupational title can be classified The following classifications were constructed A The Wright class structure Wright III B Erikson Goldthorpe and Portocarero s Comparative Analysis of Social Mobil ity in Industrial Nations schema CASMIN The European Socio economic Classification ESeC The Swiss Socio Professional Categories CSP CH Treiman s Prestige Scale The Cambridge Social Interaction and Stratification Scale Camsis 1moo For a comprehensive description of the different classifications we refer to Bergman and Joye 2001 which can be downloaded from www swisspanel ch under
102. we avoid interpreting evo lution of the sample as being an effect of attrition For numerical variables a 95 confidence interval is placed around the mean calcu lated on the set of respondents in sL and subsequently on sL for 00 08 If the interval calculated on sL does not intersect the interval calculated on sL the variable is considered to have developed a bias due to attrition in the year 20 If such a bias is detected in any year then the variable is considered to have been marked by attrition bias 23 For categorical variables the 95 intervals are placed around the percentages for each category having more than 30 respondents If the interval does not intersect for any categorical value then the variable is considered to have attrition bias for the year in question The test is first performed using the samples un weighted The process was then re peated using the cross sectional weight for 1999 and the first panel longitudinal weight for the other years to see if the longitudinal weights corrected for the bias Results There were 308 variables tested for attrition bias at each wave The variables not tested were left out due to either to much modality insufficient cases in the categories or the complete absence of variation in the 1999 responses that is all of the applicable re sponses were the same Table 1 lists the magnitude of variables apparently touched by attrition bias in any of the wave
103. weight which is the inverse of the probability of the household being selected into the sample The design weights are then adjusted due to non response in the first wave The weight is adjusted for non response up to the level of which households are fol 27 lowed for the second wave on For example households in the second panel were fol lowed if they completed at least the grid in the first year whereas for SHP only the so called complete households were followed As the rules of following varied initially be tween the two panels so do the initial weights 4 2 2a The initial weights for SHP 1 The SHP follows the complete households from wave 1 A household is considered complete if the household has responded to the grid questionnaire the household ques tionnaire and at least one person has responded to the individual questionnaire As a result the initial weight for the SHP is the design weight adjusted for non response at these three levels of response One should note that the adjustments for non response were not done by segmentation but by logit models The formula for this initial weight is given by 1 Ti Dy P PTT The inverse of the factor 7 is the design weight it is equal to the number of addresses selected in the region divided by the total number of addresses in region As a conse quence it varies from one great region to the next The p is the number of addresses that responded to the grid divided by t
104. y the following conventions were adapted Year related variables _yydnn Non year related variables individual number sex _dnn Where _ depends on the level of information P Person H Household G Grid X Proxy Where yy denotes the year 99 1999 00 2000 01 2001 Where d denotes the domain a Hobbies leisure free time lifestyle holidays etc Biography Health constitution Demographic variables Education Family climate relationships work repartition Grid Housing income financial situation and living condition variables J Qa To 20 life events geographical mobility social networks social origin Politics alo o S Religion values aspirations other than politic labour force work social status lt 40 y Violence yth Youth z other variables Where nn is a two digits number which refers to the number of the question normally the position in a block dedicated to a specific topic Two examples ar of interview maker of variable here 99 here O41 De 4 P99d01 4 Source of Quesiionnare domain there Personal individual here demnogractiy ar Of innen marker of varie here 00 here 171 he H00h17 A source of Quesiionnare domain theres Household ihare housing Constructed variables do not follow the convention of variable naming and codification These variables have a name corresponding to their contents for example wstatO0 for working statu
105. y Technical and Apprenticeship Compulsory Professional Education or Less Top Executives 1 top executives Self Employed 2 liberal 3 self employed professions Wage Earners 4 intellectuals 5 middle skilled 8 unskilled and managers employees 6 non manual 7 manual The significance of an educational attainment may vary according to the details and title of an occupation which has been taken into account in this schema For example a par ticular employee could be classified as being part of the intellectual professions based on her degree of managerial responsibility without necessarily having a university edu cation E Treiman s Prestige Scale Treiman proposes a very general stratification model for modern complex socie ties based on occupational prestige ratings that are supposedly independent of locality and invariant to national social and cultural settings His work in this area culminates in the construction and validation of the Standard International Occupational Prestige Scale Using the four nested levels of the International Standard Classification of Occupations ISCO Treiman s occupational prestige scores for each occupation within an ISCO level are averaged to produce a score for occupational groups as summarized by ISCO The subjectively attributed prestige of a specific occupation is a linked to the privilege and power which individuals enjoy based on their occupational titles b invariant across soci
106. y by means of a newsletter en closed with the advance letter at the start of each fieldwork phase In 2009 the SHP used tailored leaflets designed for specific groups of households fami lies with children couples without children people living alone and people of 65 years and older The leaflets treat topics that inform targeted households about study results that are most likely of interest to them 19 The newsletters of the years 1999 to 2008 can be viewed here http www swisspanel ch informations household php lang de amp pid 33 The tailored leaflets for 2009 here http www swisspanel ch informations results php lang de amp pid 203 3 3 5 Minimizing noncontact To minimize noncontact respondents are asked to leave their mobile number and or their e mail address If respondents are not willing to give such information or do not have a mobile number or e mail address they are asked to leave the address of an aux iliary e g a family member living outside of the household or a close friend who can help in case of losing track of the respondent Households are called on different days of the week and on different times during the day in order to minimize noncontact A bilingual interviewer responsible for administra tion and tracking of the addresses is briefed specially on how to find relocated respon dents Measures that are taken in case the advance letter is returned to sender are the follow ing Checking whether p

User Guide

Contents

Download Pdf Manuals

Related Search

Related Contents