SN 4427 - Families and Children Study
Contents
1. In 2003 2004 and 2006 2008 all children aged 11 to 15 in the family were invited to complete a short self completion questionnaire The age band was chosen to ensure the widest possible age group for whom the questions would be appropriate The children s self completion questionnaire covers leisure time activities computer access social participation sport and organised activities use of local amenities and attitudes to neighbourhood alcohol use smoking illegal drug use self esteem health and happiness attitudes to school and schoolwork relationship with parents and income Changes to the questionnaire content Although the questionnaire content of FACS does not vary greatly from one wave to the next there are instances when a particular suite of questions are added or removed from the survey This is often in response to a policy change or change in policy interest The main changes in the questionnaire by year of change are detailed below Wave 2 2000 The questionnaire was largely based on that used in 1999 The main area of change was a much expanded section on Working Families Tax Credit WFTC which replaced Family Credit in October 1999 The section on child care was also significantly modified to permit the collection of data about each child and each type of child care arrangement including hours and costs This reflected policy interest in childcare which had in turn been translated into the WFTC reforms and2. 2000 4 720 2 600 2001 family level 8 057 2 567 Longer questionnaire and more detailed household grid 2001 child level 14 822 125 2002 family level 7 878 2 420 For this and subsequent waves repeating child level sections moved to new file 2002 child level 13 966 346 2003 family level 7 740 3 089 2003 child level 13 716 543 2004 family level 7 469 3 253 2004 child level 12 947 1 019 2005 family level 7 656 3 045 2005 child level 12 831 861 2006 family level 7 461 3 417 2006 child level 12 693 943 2007 family level 5 818 2 965 2007 child level 10 632 922 2008 family level 5 888 2 946 2008 child level 10 738 923 Weights There are separate panel weights covering the original 1999 sample wLWOF and covering the 2001 sample of all families wLWAF differential attrition Cross sectional weights in 1999 and 2000 wXSW build an approximate cross section for each year in 1999 amp 2000 the cross section is for lone parents and lower income couples from 2001 onwards for all families with dependent children From 2001 onwards grossing weights WGROSSW and wGROSSP weight the sample back to known national parameters number of children age of Child Benefit recipients etc and supersede the cross sectional weights for these years See Section 6 for more details Paired transition weights trnwtWW are provided on datasets from 2002 2008 These are designed to deal with 2 Sample design and response This
3. Surestart Children and Young People s Unit Office of the Deputy Prime Minister and Department for Transport Wave 8 2006 The sample remained based on all families with children As at Wave 7 families without dependent children in the last wave were dropped from the sample Waves 9 and 10 2007 amp 2008 The sample remained based on all families with children As at Wave 8 families without dependent children in the last wave were dropped from the sample However the issued sample size at waves 9 and 10 had to be reduced and sample selection was based firstly on five priority criteria Cases not classified as a priority case were then randomly selected from the remaining eligible sample The five priority criteria are lone parents families with an equivalised income 70 below the median families with at least one disabled adult or child resident families with a living absent parent and large families defined as those with 3 or more dependent children Response rates With the exception of the first wave of FACS in 1999 the overall response rate to the surveys has remained relatively constant Table 2 shows that around four fifths of eligible families have provided a productive interview at each wave The panel response rate over FACS has remained consistently between 83 per cent and 88 per cent The booster response rate has fluctuated between 50 per cent in Wave 2 and 65 per cent in Wave 7 The FACS sample also contains a
4. 2001 2008 shows the variable names and labels and the years the variables are present in 25 Income related DVs Please note that in Waves 3 to 10 2001 2008 these DVs include income from self employed respondents partners This was not the case in 1999 and 2000 Treatment for disclosiveness The data archived in 2010 has been treated to minimise the potential for disclosiveness Treatment followed the approach adopted by the Family Resources Survey and where necessary was expanded upon i e where FACS contained variables that the FRS did not have but it was deemed necessary to treat included 26 A few very large households were deleted with more than 9 members or 8 dependent children ALL variables relating to ANY KIND of income and all variables that include council tax and or council tax benefit amounts were rounded to the nearest pound Certain variables based on earnings and related parts were top coded to ten times the mean of the value for each earnings related variable component variables were then set to zero for any top coded cases and DVs using these earnings variables were re run Age variables were topcoded to 80 The number of bedrooms variable was topcoded to 6 Certain variables were deleted those containing local area information such as Local Authority code and deprivation variables variables containing the month a divorce was completed and variables containing the property s
5. Pensions DWP and has been sponsored by several different Departments over the years including Her Majesty s Revenue and Customs HMRC Department for Education and Skills DfES Office of the Deputy Prime Minister ODPM and Department for Transport DfT The main objectives of the FACS surveys are to provide information on The effects of work incentive measures The effects of policy on families living standards and Changes in family circumstances over time Specifically to look at The impact of benefits and tax credits in supporting families with young children and Barriers to work particularly for low income families and measures to overcome such barriers The main themes covered in the FACS interview are Characteristics of families with children Health and longstanding illness disability of mothers and children Schooling travel performance in major subjects and behavioural problems Children s physical activity socialising problems with drink drugs and alcohol and mothers awareness of the Connexions service Use and opinions of local services for children and young people and mothers satisfaction with local area Education and training qualifications within families Levels of employment within families and mothers barriers in returning to work Overall level and source of families income In work support for families through Working Tax Credit and Child Tax Credit Families type and value of be
6. Work and Pensions Research Report No 161 London Corporate Document Services 36 McKay S 2003 Working Families Tax Credit 2001 DWP Research Report No 180 CDS Leeds McKay S and Atkinson A 2007 Disability and Caring among Families with Children DWP Research Report No 460 Leeds Corporate Document Services Meadows P and Rogger D 2005 Low income homeowners in Britain descriptive analysis DWP Research Report No 251 Leeds Corporate Document Services Paull G 2007 Partnership Transisitons and Mothers Employment DWP Research Report No 452 Leeds Corporate Document Services Vegeris S and McKay S 2002 Low moderate income families in Britain changes in living standards Department for Work and Pensions Research Report No 164 London Corporate Document Services Vegeris S and Perry J 2003 Families and children 2001 Living standards and the children DWP Research Report No 190 CDS Leeds Willitts M Anderson T Tait C and Williams G 2005 Children in Britain Findings from the 2003 Families and Children Study FACS DWP Research Report No 249 Leeds Corporate Document Services In house report series DWP Finch N Kemp P 2004 The use of the Social Fund by Families with Children DWP In House Report No 139 McKay S 2002 Low moderate income families in Britain work Working Families Tax Credit and childcare in 2000 Department for Work and Pensions Research Repor
7. around one third however so a separate shorter partner proxy questionnaire is often used to fill any information gaps For the 2007 and 2008 surveys no interviews were conducted with partners so data on partners for these years comes solely from the partner proxy questionnaire For the first time in 2003 children in the household aged 11 to 15 were also surveyed through a self completion questionnaire FACS Waves 1 and 2 were surveyed in the summer while the remaining surveys have taken place in the autumn of each year This means that estimates of change for 2000 2001 are based on more than one year NatCen produce technical reports each year on the conduct of the survey Datasets Data files are flat files There is a single file at respondent level for each wave There is also a file at child level containing questions asked to and about each child in the household SERIALNO is the unique linking identifier plus wPERSNO for the child level file Variable names are prefixed by a letter corresponding to the year 1999 A 2000 B etc with prefix P for partner eg hPAGE is age of partner in 2006 dataset The size of data files is indicated below All partner variables start wP but not all variables starting wP are partner variables eg wPARENT is the number of parents of the respondent in the household Table 1 Number of cases and variables in each FACS dataset Year Cases Variables Notes 1999 4 659 2 800
8. flag the check for further consideration by the researchers Researchers attempt to validate the extreme value or inconsistency by examining other characteristics of the case to see whether the keyed response could be valid For example if very high earnings are identified an attempt is made to examine whether this is caused through the payment of a bonus or due to the respondent concerned being in a high wage occupation If a satisfactory explanation for the check is not forthcoming from either of these two processes then previous years data for cases with similar characteristics to the case reporting the extreme value are examined If it has not be possible to resolve the inconsistency or extreme value at this stage the value remains in the data file If it is a value associated with a benefit the presence of an inconsistent or extreme value is noted in the benflag variable Once the editing and coding of the FACS data was completed initial datasets at both family and child level were prepared Numerous additional checks were carried out on all variables in these datasets by NatCen and external collaborators specifically Frequencies of all variables in each new waves dataset were run and the results were compared with those from previous waves to check consistency In addition the means medians and maximum and minimum values of all continuous variables were compared with those from previous waves DWP also examined the initial
9. from survey respondents The other major change was the introduction of a self completion for children aged 11 15 This aimed to covered their attitudes to school leisure activities and interests behaviour and so on The questionnaire also retained a large series of questions about children that were directed at the main respondent usually the mother Wave 6 2004 The main change to the questionnaire was in the section about Tax Credits Additional questions were added to this section to collect information about the annual Tax Credit renewal process The self completion for children aged 11 15 was retained at Wave 6 and the content remained the same as at Wave 5 Wave 7 2005 Firstly in previous waves most of the questions from the childcare section were asked to working families only In Wave 7 data on childcare has been included for working and non working families Secondly the self completion questionnaire for children aged 11 15 was dropped Finally a new section on Social Capital for main respondents and partners has been included Wave 8 2006 Three changes to the FACS interview occurred in Wave 8 Firstly the self completion questionnaire for children aged 11 15 was re introduced Secondly questions about expenditure hardship and job readiness were dropped Finally sections on attitudes self completion section and future plans were added These sections included a task requiring the respondent to sort small show c
10. its component element the childcare tax credit Wave 3 2001 A new module on child outcomes was included in the questionnaire Wave 4 2002 The questionnaire was able to include more questions about children Sections relating to education were expanded A whole new set of questions related to the use of services and contact with various professionals involved in family and child related services such as teachers social workers doctors etc Parents were also asked about how their children travelled to school each day and their reasons for different choices of transport Children s social lives were also a subject of inquiry for the first time This increase in the number of questions about children was compensated for by scaling back questions relating to WFTC which by then was in its last six months The self completion was also modified to include questions about schooling 20 Wave 5 2003 Two main changes were made to the questionnaire New Tax Credits Working Tax Credit and Child Tax Credit were introduced from April 2003 and administered by the Inland Revenue This necessitated a major new series of questions covering all aspects of the reform Changes were also needed to the way that income data was collected in that the New Tax Credits are generally paid at a fixed level for a year and later reassessed on the basis of income within that year Therefore greater efforts were made to obtain relevant documentation
11. or imputed However the quality of self employment income data does not appear to be greatly affected by the lack of editing an investigation of 2007 data showed very few cases where an edit would have been triggered Attitudinal data Take care when using the attitude variables those with P inserted after the wave specific prefix are from partners self completion questionnaires 31 e g All things considered how happy are you wPQ1SC for partners wQ1SC for main respondents Benefits Benefit receipt DVs are typically just the name e g wWFTC wilS Benefit amount DVs are the name plus e g wCTB wIB Housing benefit is not included in the benefits section Housing Benefit receipt is indicated by the variable wHOUS27 If Housing Benefit receipt is required as a proportion of all recipients it needs to be recoded as this question was filtered and only certain respondents answered it For example compute WHB 0 If WHOUS27 eq 1 wHB 1 SOC SIC codes Derived variables for SIC and SOC codes Nine categories for SOC groups 1 Managers and senior officials 2 Professional 3 Associate professional technical 4 Administrative secretarial 5 Skilled trades 6 Personal services 7 Sales customer services 8 Process plant machine operatives 9 elementary And nine categories for SIC codes 1 Agriculture forestry and fishing 2 Mining and quarrying 3 Man
12. regardless of any changes in their circumstances i e there was no screening exercise movers were followed even if outside the originally sampled areas The second element aimed to represent a new cross section of lone parents and low income couple families This was achieved through re screening the families found to be ineligible at Wave 1 and screening new Child Benefit recipients in the selected areas specifically new families and in movers to those areas Fieldwork took place during June to early October 2000 The questionnaire contained new questions on the Working Families Tax Credit WFTC The interest in collecting information on WFTC meant that the screening stage was modified in two ways First the level of income cut off was raised significantly to include any families potentially eligible for WFTC Secondly those families that paid for childcare were routed through an in home screener to ensure that the potential value of the Childcare Tax Credit was included in the income screener Rates of WFTC were equalised for younger children so the 1999 CAPI Computer Assisted Personal Interviewing screener which had to be based on different rates of Family Credit FC for children of different ages did not need to be used in year 2000 As an incentive for taking part those households participating for the second full interview received a 10 Boots voucher irrespective of family composition This practice has been maintained in su
13. the boosters plus panel families are self weighting to a contemporary cross section Attrition weights used for the panel weights of 1 for that years boosters 29 7 Advice for users This chapter includes some advice on using FACS data and some warnings about inconsistencies in the datasets Household grid Household grid information is contained at the start or near the start of each dataset immediately after any unique identifier information The household grid provides details of each member of the household including their age gender and current employment status but most importantly their relationship to other members of the household space is available to detail up to fifteen family members although this was restricted to nine slots prior to 2001 e g wRELR_02 details the relationship of person 2 in the household grid to the main respondent It should be noted that each person retains their unique position in the household grid from one wave to the next the main respondent is always given the first position in the household grid However this position can change if the original main respondent leaves the household and another family member new or existing becomes the main respondent the variable WRPERSNO advises what position the main respondent locates SPSS syntax tips Matching Files Individual wave datasets at the family level can be easily merged using the Match Files command in SPSS and using SE
14. FACS datasets containing raw variables only and compared key estimates from previous waves data to current wave raw variables to ensure constancy 15 Steve McKay at the University of Birmingham compared key household demographic variables income and benefit variables raw variables and derived variables from wave to wave comparing the distribution of answers numbers of missing values size of max min HMRC compared the Wave 7 Tax Credit claimant figures with administrative data to ensure the estimates in FACS were consistent with official data 16 4 The data collection process and survey content This chapter covers a brief description on the fieldwork and most importantly the structure and content of interviews questionnaires and self completion questionnaires FACS respondents The FACS sample is a named respondent sample with names and addresses selected at random from Child Benefit records Because the aim of FACS is to collect information on families with children the named respondent acts as the main means of making contact with the family and for collecting information on the various aspects of family life To ensure the circumstances of families with children are fully captured the study is designed to collect various types of information on the family as a whole and separately for the main respondent usually the family s mother figure the person with the main responsibility for looking after the ch
15. Kay McKay 2003 Many thanks for his permission to include parts of the earlier document note for this report As discussed in Section 1 Child Benefit records were used as the sampling frame for the initial sample Wave 1 1999 The sample was selected from the then DSS Child Benefit records The sample was initially stratified by Region and a rural urban measure within region and a sample of 150 postcode districts or clusters from the national total of 2 600 districts was selected proportionate to the numbers of Child Benefit records in each district An equal number of records identified in each of these postcodes was then selected every nth record starting from a random point within each sector The programming routine was set to produce 100 families in each of the 150 sectors to provide a starting sample of 15 000 families with children This was before any opt outs or screening or removal of invalid addresses etc A separate booster sample of Family Credit recipients was drawn in 1999 comprising both a stock and inflow sample These were followed up in the next two years of FACS but then dropped They appear only sporadically in published reports and have been removed from the public use datasets Wave 2 2000 The Wave 2 sample had two main elements which are relevant for all subsequent waves Firstly the panel element contained all families from the 1999 survey who were eligible for re interview respondents and any partners
16. London National Centre for Social Research Woodland S and Collins D 2000 Study of families with children technical report London National Centre for Social Research Woodland S and Woodward C 2002 Families and Children Study FACS 2000 Wave 2 technical report London National Centre for Social Research Woodland S et al 2003 Families and Children Study FACS 2001 Wave 3 technical report London National Centre for Social Research Other Publications Barnes M Lyon N and Conolly A 2006 The Living Standards of Children in Bad Housing Evidence from the Families and Children Study FACS 2004 London National Centre for Social Research Emerson E and Hatton C 2005 The Socio Economic Circumstances of Families Supporting a Child at Risk of Disability in Britain in 2002 Lancaster Lancaster University Haux T 2007 Lone parents with older children and welfare reform CASP Working Paper University of Bath Available at http www policyhub gov uk http www bath ac uk casp assets LoneParentst pdf Levitas R Pantazis C Fahmy E Gordan D Lloyd E and Patsios D 2007 The Multi Dimensional Analysis of Social Exclusion Cabinet Office London Government Chief Social Researcher s Office Available at http www cabinetoffice gov uk social_ exclusion task force documents Morris S 2007 Mothers child support arrangements a comparison of routes through which mothers obtain award
17. RIALNO as the key variable on which to match The SPSS syntax below creates a ten wave merged dataset although given the size it is wise to extract a reduced number of variables before conducting a merge to create a wide panel dataset I ACTION Substitute relevant local directory names here FILE HANDLE F1 NAME C FACSCD2 Data facs1999a sav FILE HANDLE F2 NAME C FACSCD2 Data facs2000a sav FILE HANDLE F3 NAME C FACSCDa2 Data facs2001a sav FILE HANDLE F4 NAME C FACSCD2 Data facs2002a sav FILE HANDLE F5 30 NAME C FACSCD2 Data facs2003a sav FILE HANDLE F6 NAME C FACSCDa2 Data facs2004a sav FILE HANDLE F7 NAME C FACSCD2 Data facs2005a sav FILE HANDLE F8 NAME C FACSCD2 Data facs2006a sav FILE HANDLE F9 NAME C FACSCD2 Data facs2007a sav FILE HANDLE F10 NAME C FACSCD2 Data facs2008a sav This will merge together the ten waves of data MATCH FILES file F1 file F2 file F3 file F4 file F5 file F6 file F7 file F8 file F9 file F10 by serialno Show some of the overall interview history freq var AINT BINT CINT DINT EINT FINT GINT HINT IINT JINT Should get the numbers of valid cases interviewed in each year as shown in table 1 above Warnings Income data Income from self employment is not included in the 1999 and 2000 datasets Note also that self employed income data unlike data on employment and benefit income has not been edited
18. UK Data Archive Study Number 4427 Families and Children Study Families and Children Study FACS Waves 1 to 10 User Guide 1999 2008 Prepared by NatCen for DWP Dan Philo Lorenc Hoxhallari Anne Conolly and Matt Barnes 1 IAtFOCUCTION 0 oe eee cecceecceccecceeccecceececceccueecuccueeaueceecaeeseeceeaeeceecaesauecuecaesseeceeeaeeceesteeaeeceeens 3 The Families and Children Study FACS cccceeeeeeeeeeeeneeeeeeeeeeeeeseeeaeeeeeeeeeeeeenees 3 The purpose of this user QUICC cccceeeeeeeeeeeeeeeeeeeeeeeeeeeeeaaaeeeeeeeeeeeeseneaaeeeeeeeeeeeeenee 4 PEGUICK Start QUIN pact is co aa Gab nani oe Pek eo a tata tan i een tet tate in eal 5 2 Sample design and response ccccecccceeeeeeeeeeeeeeeeeeeeeeeeeeeseeaaaeaeeeeeeeeeeeesceneeaeeeeeees 7 Sai ple dESiIgM eis on R RA a 7 Response rales are raae aiaee eara Eaa ee Salar aE aaan hee aaa E Eaa ARE ae ante Ante 11 Attrition eon a aAa Na Eaa E Aa a AE E Aa AAA AEA n TAEA A NA EAA 11 3 Preparing th dataSetinnpoieni aa a a aaa 14 The edit PV OCESS ekini idi a i aai aii aii aa 14 NPL AUO Mes akon ete cna e a e a EaR be Ea aah a che a bed 14 F rther data CNOC kS ic 2 a a eons wade es ede a a Aa aaa iaaa ea adii 15 4 The data collection process and survey content essssssseerrerssesrrrrrnrreneserrrrerenn 17 FACS respondents xt s ci2 deer te nas ek beh esas ahaha tee elas Ne etal at 17 The content of the interviews 2 cccccc
19. a dependent child even if at the time of interview they had finished their course of study The content of the interviews The survey fieldwork has been conducted by the National Centre for Social Research NatCen The FACS surveys are carried out via a face to face interview with the mother and for the 1999 2006 surveys the partner if present In some years parents or children complete a self completion interview The structure of the interview was as follows A one hour Computer Assisted Personal Interview CAPI with the mother For couple families in 1999 2006 surveys A 15 minute CAPI interview with the partner if the partner is not present a short proxy partner interview with the mother is carried out A 10 minute self completion questionnaire depending on the year of the study administered to parents or children aged 11 to 15 years Mother s interview The main themes covered in the mother s interview are Information about the family unit Family composition Relationship histories Contact with non resident parents Social Capital Housing Receipt of other social security benefits Attitudes to work and family life and future plans 18 Receipt of the New Tax Credits Working Tax Credit and Child Tax Credit questions on the New Tax Credits replaced questions on Working Families Tax Credit and Children s Tax Credit which have been replaced by the New Tax Credits Other income and savings a
20. al 1999 ordinai of last 2000 oraiial of last 12001 original of last 2002 y g wave g ave g ave cohort cohort cohort 1999 4659 100 100 T 2000 3560 76 75 1160 100 25 2001 2002 8031 65 38 2003 2651 57 1384 2004 2313 50 81 2005 2135 46 28 606 52 8 2251 58 29 2006 1876 40 25 541 47 1 1932 50 26 AS A AS of last 2003 _ __ lof last Bue original NE cohort C eS ie a ee See eae e a ee 3 8 515 67 1 Year of first interview As FACS of year original of last 2005 cohort 2004 803 100 11 2005 668 83 l9 2006 603 76 8 645 83 9 771 100 10 Excludes family credit sample 13 3 Preparing the dataset This chapter briefly describes the process of getting the data from the survey instrument the CAPI instrument to a final dataset This process involves editing the data imputing missing values and a string of data checks Each of these stages of the data preparation process is discussed in more detail in the survey technical reports The edit process Interviewers carry out most of the data validation of CAPI surveys in the field Interviewer checks in the CAPI program allow interviewers to clarify and query any data discrepancies directly with the respondent The CAPI program applies range and consistency error checks and both types of checks were used extensively throughou
21. ards in order to provide information as well as answering standard interviewer CAPI questions Wave 9 2007 The questionnaire remained the same as for wave 8 apart from the addition of three questions about limiting long term illness asked of the respondent about themselves and each child 21 Wave 10 2008 The questionnaire remained the same as for wave 9 apart from very minor changes e g interviewer prompts 22 5 The FACS datasets This chapter introduces the user to the FACS datasets It covers issues such as variable naming conventions key linking variables and derived variables All the datasets and syntax provided are in SPSS format and this chapter uses SPSS terminology throughout Accessing the data The FACS datasets are deposited at the Data Archive at Essex University The data and documentation can be downloaded from this website See http www data archive ac uk for more details The datasets Levels of data There are two datasets created for each wave of FACS e a family level dataset which contains information about the family the mother and her partner if present wWPARTNER indicates whether there is a partner in the household or not and e achild level dataset which contains information about each separate child in the household Child level variables are prefixed by a c after any wave identifier The children within each household are assigned a number which appears at the end of the va
22. bsequent years with a switch to the more generic and widely useable High Street Vouchers Wave 3 2001 In 2001 the study was expanded to include families of all income levels Lone parents have been included from the outset whatever their level of income Hence the study now included middle and higher income couples who had been excluded from previous waves The panel element of the study attempted to re interview respondents and any partners of all families regardless of any changes in circumstances The cross sectional element of the study attempted to achieve a new representative sample of all families by interviewing families found to be ineligible at Wave 1 and or Wave 2 on grounds of high earnings There were also attempts to top up the sample with new Child Benefit recipients in the selected areas specifically new families and in movers to those areas Fieldwork for Wave 3 was moved from Summer to Autumn There were several reasons for this This shift permitted greater investment of time in programming and testing the questionnaire needed because the survey expansion generated a significant new module on child outcomes An additional motivation was to ensure that WFTC recipients had all qualified under the same annual regime uprating in April occurring about 6 months before proposed fieldwork Last it circumvented certain ambiguities about the economic status of school leavers and those aged 16 18 that could arise when fieldwor
23. cceeeeeeeeeeeeeeaeeeeeeeeeeeeeeeeeaaaaeeeeeeeeeeeseescnaaaaeeeeeees 18 Changes to the questionnaire CONTENL ccceeeeeeeeee eee eeeeeeeeaeeeeeeeeeeeeeeeeeneeeeeeeeeeeeeeeed 20 Be WG EACS CalaSels eect es sce vee teid ces ess aued onc ened aes swed E 23 Accessing ING datai eats tee ake t Nees he eee Need See heed Ae eee ees 23 WAG OLAS Ch Sacto Ae etearaseet N a Aa N ane aa aa 23 Derived Variables DVS jects ciictataiticen eli nta n A E 25 M2 Weighting ssns eee re ge cate eee edd Set cand aS cede acess eee 27 Using the FACS weights casa Aaa A aR aah a ae tae ia oe oh Steg 28 TUNOWICG TORUST take a sed ee Nee hee hed ees eee ted Nes Ne ee ted ee Nec eel haa ae Nee 30 SPSS SyNtaX TOS ei a ciate let eat abencpte te tebe aa a a a A 30 MV GRINS a Ainea tes ee ae eee sees 31 8 Further Information casi cect ticti eet cacti ete ete eit etl teases ete tates 35 Published FACS 1 OQ ONS ccc fcosc ccd cemcceiemenstd sane sege coun dasesencdexexenddesatedddovanendtusesenieadonete 35 1 Introduction The Families and Children Study FACS The Families and Children Study FACS is a series of annual panel surveys designed to investigate the circumstances of British families with dependent children The study began in 1999 with a survey of all lone parent families and low moderate income couples In 2001 the third annual study was enlarged to be representative of all families with dependent children The study is commissioned by the Department for Work and
24. chapter covers technical information on sample design and response rates Sample design The dual objectives of the initial sample design were to provide a representative sample of Britain s low income families while at the same time generating a sample of sufficient size for a longitudinal study Sampling procedures for later years of the study were generally based on the previous year s outcomes Interviews are sought at each wave with productive cases at the previous wave and where permission had been given to be re contacted Each year the longitudinal sample is refreshed with a booster sample of new families in order to ensure it is representative of all families The booster sample is made up of new families where there has been the birth of a baby and in movers families new to the sampled postcode areas This general sample design is illustrated in Figure 2 1 Figure 2 1 FACS panel conceptual sample design Year N Year N 1 Non response Main families Panel sample No dependent F New families children Leaving sample area In movers STOP START Where possible considerable efforts are made to retain those families in the panel This ensures that FACS allows for meaningful longitudinal analysis There have been a number of wave specific changes to the sample design which are detailed below 3 This section draws upon an earlier document written by Steve Mc
25. council tax band 6 Weighting This chapter provides a brief overview of the weights supplied on the FACS datasets The chapter also includes some practical advice on how to use the weights The FACS datasets include a number of different weights Grossing weights 2001 2008 datasets only A set of grossing weights has been derived The aim of the grossing weights is to make the FACS sample look like known national figures for the distribution of families with children in that year The grossing weights are e wGROSSW a grossing weight to survey numbers with an average of 1 e wGROSSP a grossing weight to population numbers calculated by multiplying wGROSSW by a constant to give around seven million families receiving Child Benefit Cross sectional weights 1999 2000 datasets only Note that these weights are superseded by the grossing weights from wave 3 onwards A set of cross sectional weights wXSW builds upon the panel weights and includes the booster samples with weight 1 to produce a weighted sample close to what would be produced from a contemporary cross section survey Longitudinal weights A set of longitudinal weights have been designed to allow users to conduct weighted longitudinal analysis The balanced panel longitudinal weights are e wLWAF used to weight the 2001 sample for attrition between 2001 and 2008 It takes the 2001 grossed sample of families as its base For those interested in all families
26. d Lyon N 2007 Families with Children in Britain findings from the 2005 Families and Children Study FACS DWP Research Report No 424 Leeds Corporate Document Services Kasparova D Marsh A Vegeris S and Perry J 2003 Families and children 2001 Work and childcare DWP Research Report No 191 CDS Leeds Kempson E McKay S and Willitts M 2004 Characteristics of families in debt and the nature of indebtedness DWP Research Report No 211 Leeds Corporate Document Services Lacovou M and Berthoud R 2006 The economic position of large families DWP Research Report No 358 Leeds Corporate Document Services Lyon N Barnes M and Sweiry D 2006 Families with children in Britain Findings from the 2004 Families and Children Study FACS DWP Research Report No 340 Leeds Corporate Document Services Marsh A et al 2001 Low income families in Britain work welfare and social security in 1999 Department of Social Security Research Report No 138 London Corporate Document Services Marsh A and Rowlingson K 2002 Low moderate income families in Britain changes in 1999 2000 Department for Work and Pensions Research Report No 165 London Corporate Document Services Marsh A and Perry J 2003 Family Change 1999 to 2001 DWP Research Report No 181 CDS Leeds McKay S 2002 Low moderate income families in Britain work Working Families Tax Credit and childcare in 2000 Department for
27. e child level 33 Proxy data Proxy data for partner interviews has been included in the family level data file Users are advised to avoid using proxy data in cases where actual partner interview data has been obtained For the 2007 and 2008 surveys no interviews were conducted with partners so data on partners for this year comes solely from the partner proxy questionnaire The effect of this was found to be minimal This was examined by Steve McKay and his report is included in Conolly et al 2009 34 8 Further Information This chapter contains information on documents that contain extra information on FACS and published reports based on analysis of the FACS data Technical reports Technical reports for each wave of the survey are available from the National Centre for Social Research http Awww natcen ac uk study families and children study facs FACS user website http www dwp gov uk asd asd5 facs FACS respondent website http www natcen ac uk facs Published FACS reports Atkinson A Finney A and McKay S 2007 Health Disability Caring and Employment Longitudinal Analysis DWP Research Report No 461 Leeds Corporate Document Services Barnes M Willitts M et al 2004 Families and Children in Britain Findings from the 2002 Families and Children Study FACS DWP Research Report No 206 CDS Leeds Barnes M Lyon N Morris S Robinson V and Yau Y 2005 Family Life in Britain Findings
28. from the 2003 Families and Children Study FACS DWP Research Report No 250 Leeds Corporate Document Services Barnes M Connolly A and Tomaszewski T 2008 The Cirumstances of Persistently Poor Families Evidence from Families and Children Study DWP Research Report No 487 Leeds Corporate Document Services Berthoud R Bryan M and Bardasi E 2004 The dynamics of deprivation the relationship between income and material deprivation over time DWP Research Report No 219 Leeds Corporate Document Services Berthoud R and Blekesaune M 2007 Persistent employment disadvantage DWP Research Report No 416 Leeds Corporate Document Services Brewer M and Paull G 2005 Newborns and new schools critical times in women s employment DWP Research Report No 308 Leeds Corporate Document Services 35 Browne J and Paull G 2010 Parent s Work Entry Progression and Retention and Child Poverty DWP Research Report No 626 Leeds Corporate Document Services Evans M Harkness S and Ortiz R 2004 Lone parents cycling between work and benefits DWP Research Report No 217 Leeds Corporate Document Services Farrell C and O Connor W 2003 Low income Families and Household Spending DWP Research Report No 192 CDS Leeds Hedges A 2001 Moving between sickness and work Department for Work and Pensions Research Report No 151 London Corporate Document Services Hoxhallari L Conolly A an
29. ildren in the family the partner if present and any dependent children A definition of each of these units of interest is provided below The family unit In FACS the family unit must comprise at least one dependent child see definition of dependent child below and at least one adult who is responsible for this child The adult responsible for the child can be the child s natural or adoptive parent or the legal guardian s to whom Child Benefit is paid The definitions used in FACS means that families cannot span more than two generations so for example grandparents and grandchildren living together are not considered to belong to the same family The exception to this however is where the grandparents are responsible for looking after the grandchildren where the parents are not part of the family unit The main respondent mother figure In FACS information about the family is collected principally from one family member usually the mother or the mother figure Often the main respondent is the natural or adoptive mother of the dependent child ren who lives with her However in some cases she could be the cohabiting partner of the child ren s natural or adoptive father a grandmother or other female guardian The intention is for father figures to be interviewed as the main respondent only in cases where there is no female mother figure present in the family In the 2005 study only two per cent of main responden
30. k occurred during the summer months Wave 4 2002 By Wave 4 the FACS sample remained relatively stable including all families with children However a range of other Government sponsors joined the study bringing with them even more of a focus on the lives of children The Family Credit boosters were dropped from the sample in this wave Wave 5 2003 The sample remained based on all families with children Prior to Wave 5 families that have ceased to have dependent children over the course of the survey continued to be interviewed for the next few waves of FACS In this wave this policy of retaining those without dependent children was reviewed and the decision was taken to only interview families for one year once they had ceased to have dependent children Families without dependent children were therefore automatically dropped from the sample if they had been classified as a family without dependent children at the previous interview This rate of ageing out in FACS has been estimated at 3 5 per cent of families each year McKay 2003 Wave 6 2004 The sample remained based on all families with children As at Wave 5 families without dependent children in the last wave were dropped from the initial sample Wave 7 2005 The sample remained based on all families with children As at Wave 6 families without dependent children in the last wave were dropped from the sample Department for Education and Skills
31. nd Expenditure and hardship Information about the main respondent herself Education and training Health Caring responsibilities Employment and self employment Work history and Unemployment and job search Information about each specific dependent child Health School and education Problems and use of local services Parental aspirations for children and Childcare arrangements Partner s interview For couple families 1999 2006 a short interview was carried out with the partner The partner interview includes information on Education and training Health Employment and self employment Earnings Unemployment and job search and Caring responsibilities In cases where the partner interview is not completed and in 2007 where there was no partner interview a proxy interview is carried out with the mother This is to minimise the risk of having no data about the partner The proxy partner interview collects information on Current or recent work status Earnings Industrial and occupational classification data and Qualifications Having collected this proxy information the intention where possible is for the interviewer to return to the family at a later date to conduct an interview with the partner 19 Self completion questionnaires Waves 1 4 of the survey had a self completion questionnaire for the main respondent and their partner covering morale and various attitudinal questions
32. nefits and families on Income Support Arrangement and payment of child maintenance and contact between children and absent parents Money management savings and debt Use of formal and informal childcare and local childcare services Type of property and tenancy and quality of housing Material deprivation household and leisure items that the family is unable to afford and 1 Low moderate income couple families were defined as eligible if neither adult was working or the family income was less than 275 a week for a family with one child with this threshold increasing to a weekly income of less than 425 for a family with four or more children Attitudes to work and family life and future plans The purpose of this user guide Users should note that data improvements have been made to the eight datasets 2001 2008 For more details see Data Versions section in Chapter 5 of this guide The purpose of this user guide is to help new users become familiar with the Families and Children Study FACS data It provides an overview of the structure of the datasets basic details on editing imputation and weighting and various tips on how to use the data The user guide has eight chapters Chapter 2 contains technical information on sample design and response rates Chapter 3 includes explanations on data editing imputations and further data checks Chapter 4 provides information on the data collection process This chapter al
33. ner information Information about partners is contained within the family level datasets The variable names for information pertaining to partners are prefixed by a P although this comes after any wave identifying prefix For example WwPUSWAGE Earnings from employment in main job weekly relates to the partner whereas wWUSWAGE relates to the main respondent All partner variables start wP but not all variables starting wP are partner variables e g wPARENT is the number of parents of the respondent in the household 24 Multiple response variables All multiple response items for Waves 6 to 10 are supplied as dichotomous variables More specifically the first item detailed on a showcard will represent a variable and a value of 1 will be allocated if the interviewee mentioned that item or System missing if they did not The second item will represent the second variable and a value of 2 will be allocated if the interviewee mentioned that item or System missing if they did not and so on The number of variables supplied therefore reflects the number of items detailed on the showcard The codeframes have been made consistent across these waves to make analysis easier Multiple response items in Waves 3 to 5 that are also present in later waves are also supplied as dichotomous variables as above Multiple response items in these earlier waves that are not present in later waves have generally NOT been c
34. ns D and O Connor W 2002 Encouraging and Maintaining Participation in the Families and Children Survey Understanding why people take part DWP Working Paper No 6 Plewis And Hawkes D 2005 Feasibility study into the effects of low income material deprivation and parental employment on outcomes for children both in adulthood and as children DWP Working Paper No 20 Willitts M 2006 Measuring child poverty using material deprivation possible approaches DWP Working Paper No 28 Technical Reports published by NatCen Conolly A Maplethorpe N and D Souza J 2009 Families and Children Study FACS 2007 Wave 9 Technical Report London National Centre for Social Research Kerr J Conolly A and D Souza J 2007 Families and Children Study FACS 2006 Wave 8 Technical Report London National Centre for Social Research Lyon N Scholes S and Tait C 2005 Families and Children Study FACS 2003 Wave 5 Technical Report London National Centre for Social Research 38 Lyon N Tait C and Scholes S 2006 Families and Children Study FACS 2004 Wave 6 Technical Report London National Centre for Social Research Lyon N Mangla J Tait C and Scholes S 2007 Families and Children Study FACS 2005 Wave 7 Technical Report London National Centre for Social Research Phillips M Miers A and Scholes S 2003 Families and Children Study FACS 2002 Wave 4 technical report
35. onverted to dichotomous variables and remain in multi coded format In this case where these also have an option for open text answers the edit variables used to code these open answers are also supplied Multiple response items in Waves 1 and 2 are generally supplied in their original multi coded format along with any edit variables like with Waves 3 to 5 Imputation variables In general when imputations have been made the original variable has been overwritten but a flag variable has been created to inform the analyst how many cases have been affected by the imputation The flag variable contains the value that has been replaced following BHPS practice typically 998 in the case of a refusal and 999 for respondents who don t know the answer Imputation flags are suffixed by X For example wLASTPYX is the imputation flag for wWLASTPAY last pay weekly Missing values No values have been declared missing on the files supplied Generally speaking for variables based on the questionnaire 999 8 means not answered and 999 9 means don t know These are indicated as value labels Derived Variables DVs Each wave of data is accompanied by a core set of derived variables DVs DV are derived from the questionnaire variables to enable meaningful and easier analysis The document FACS 1999 2008 Derived variables specifications provides specifications for each DV and the excel file Table of Derived Variables
36. ords whether the earlier or later version of the CAPI questionnaire was used Childcare was asked at respondent level in 1999 but at child level in subsequent years In 2005 following a review of childcare data by Brewer and Shaw in 2003 Brewer and Shaw 2004 Childcare use and mothers employment a review of British data sources DWP Working paper 16 CDS it was recommended that the childcare section of FACS should be adapted to make it more consistent with other data sources Substantial changes were therefore made at Wave 7 2005 which restrict comparisons of childcare use in FACS over time questions relating to the perceptions of childcare remain comparable with previous waves Maintenance Maintenance was collected in 1999 as part of the relationship block taking each absent parent as the relevant unit This was thought to be unsatisfactory since there is missing data in the relationship block the CAPI instrument allows for respondents to decline the whole section since some of the questions may be thought sensitive In 2000 and for 2001 maintenance was collected at child level much earlier in the questionnaire This avoids missing data but makes it more difficult to link child maintenance data to characteristics of former partners From 2002 onwards maintenance was again collected early in the question but at respondent level This was mainly to save time given a large expansion in the number of questions asked at th
37. oss sectional weights remove families with no children automatically from the sample longitudinal weights do not Weighting the 1999 data AINT has a value of 1 for those interviewed in 1999 There are no weights for the 1999 data there are variables wLWOF and wXSW which are set to 1 for consistency Weighting the 2000 data BINT has a value of 1 for those interviewed in 2000 There is a weight variable for the 2000 data that deals with attrition during 1999 2000 bLWOF A cross sectional weight bXSW generates an approximate cross section of lone parents and low income families in 2000 For 2000 to look at the panel dataset WEIGHT by bLWOF wLWOF denotes Longitudinal Weight Original Families Weighting the 2001 data CINT has a value of 1 for those interviewed in 2001 There are three weights provided a panel weight cLWOF and two grissong weights CGROSSW and cGROSSP For 2001 to look at the cross section SELECT IF CGROSSW gt 0 28 WEIGHT BY CGROSSW Weighting the 2002 2008 data wINT has a value of 1 for those interviewed in that year dINT 2002 elNT 2003 etc There are five weights provided a panel weight wLWOF for the original 1999 families a panel weight for the 2001 sample of all families wLWAF grossing weights WGROSSW and wGROSSP and a paired transitions longitudinal weight wtrnwtww To look at the cross section The cross section is an approximation based on the idea that
38. rather than low income families Wave 3 2001 may be regarded as something of a base year in the FACS series of surveys It is updated for each wave taking 2001 as the base year e wLWOF used to re weight for attrition the families interviewed in the first wave 1999 Weights are defined only for those participating in every wave the methodology used in BHPS This set of weights is for those interested in the longer time series available for lower income couple families and for lone parents wLWOF denotes Longitudinal Weight Original Families 27 The paired transitions longitudinal weight is trnwtww used to weight for attrition and any potential non response bias for families that participated in consecutive waves For example trnwtij is for Waves 9 to 10 Using the FACS weights This section provides some general advice on using the FACS weights including weight cross sectional and longitudinal analysis Analysing families with dependent children Families that have ceased to have dependent children over the course of the survey may have still been interviewed interviewed for one year only from FACS 2002 Wave 4 but for 2007 and 2008 generally not interviewed if found to have no dependent children Therefore these families will generally need to be removed from analyses e g SELECT if wWNDEPCH gt 0 This is important when producing unweighted bases and when performing some kinds of panel analysis Whilst cr
39. rescreen sample from Waves 2 to 4 and an opt in panel sample from Waves 5 to 8 The rescreen sample response rate has varied between 36 per cent for the in wave 2 to 70 per cent in wave 3 when the rescreen sample contained those higher income families who were previously ineligible in Waves 1 and 2 Table 2 Table 2 Productive interviews at each wave of FACS by sample type Panel Booster Opt in rescreen Wave productive productive productive Overall interview interview interview oe Count al count io count er count 1 n a n a 50 4 659 n a n a 50 4 659 2 83 4 092 50 474 36 686 78 5 252 3 84 4 671 56 727 70 3 143 80 8 541 4 84 6 919 59 811 59 153 80 7 883 5 87 6 846 56 767 65 133 82 7 746 6 84 6 578 60 803 58 90 80 7 471 7 87 6 773 65 782 57 102 83 7 657 8 85 6 577 61 771 75 116 84 7 464 9 88 5 149 63 603 60 69 84 5 821 10 87 5 115 59 701 64 72 82 5 889 In Technical or Annual Reports the number of productive interviews may be different to the interview numbers reported here This is due to the removal of families with no dependent children or the weights assigned to each case Attrition The same respondents are interviewed every year in FACS as part of a panel sample However as with any other panel study e g the British Household Panel Survey BHPS there can be problems with respondents dropping out of the sample termed attrition leading to the panel becoming either too small or unrepresentative of the total popula
40. riable name Data versions The data archived in 2010 had improvements made to the datasets 2001 2008 to make them easier to use and more consistent The main improvements were e Derived variables were added so that most of these datasets contain a full set of comparable DVs e Multiple response items were changed into dichotomous variables see section Multiple response variables for more details e Self employment income was added to the income DVs and structure of the income DVs was made more user friendly e Some variable names were changed to make them consistent and or to facilitate longitudinal analysis e The variable order was changed to approximate that of the questionnaire see Variable order section below for more details e The data was treated to minimise the potential for disclosiveness see Treatment for disclosiveness section below The datasets are named FACS 1999 sav FACS 2000 sav and so on Different release versions of the data are suffixed by sequential letters of the alphabet a indicates that it is the first release of the data e g FACS 1999a sav b that it is the second release of the data e g FACS 1999b sav 23 It is important that analysts ensure that they use the most up to date versions of data supplied Different release versions for example may include corrections or updates to derivations We also re release earlier waves of data each year so that con
41. s for maintenance in Britain Benefits The Journal of Poverty and Social Justice 15 1 17 31 Spencer N J 2005 Does material disadvantage explain the increased risk of adverse health educational and behavioural outcomes among children in lone 39 parent households in Britain A cross sectional study Journal of Epidemiology and Community Health Vol 59 pp 152 7 Spencer N J 2005 Maternal education lone parenthood material hardship maternal smoking and longstanding respiratory problems in childhood testing a hierarchical conceptual framework Journal of Epidemiology and Community Health Vol 59 pp 842 6 Spencer N J 2006 Social equalisation in youth evidence from a cross sectional British survey European Journal of Public Health Vol 23 Epub ahead of print 40
42. sis The document FACS 1999 2008 Derived variables specifications provides specifications for each DV and the excel file Table of Derived Variables 2001 2008 shows the variable names and labels and the year s in which each variable is present in the dataset Showcards The full versions of the questionnaire showcards are provided Techncal reports Produced each year these contain comprehensive details of the surveys including sample design development work fieldwork response weighting and imputation coding editing and checking of data and the full suite of relevant documents A quick start guide Background The Families and Children Study FACS surveys have taken place annually since 1999 The basic design is of a panel study with sample boosters that ensure cross section representativeness in each year The 2001 sample is significant samples in 1999 2000 were only of lone parents and low income couples with children approximately the poorest 40 of couple families In 2001 the sample was extended to all families with children by including higher income families previously screened out The sample is taken from Child Benefit records so is strictly a sample of Child Benefit recipients rather than of all British based families with children Questionnaires Data is collected in almost all cases from the Child Benefit recipient and in 1999 2006 from the respondent s partner usually where present Partner non response is
43. sistent and new derivations are supplied for each wave of data Wave Identifiers The letter w prefixed before a variable name is used to represent the wave Variable names in 1999 Wave 1 are prefixed by a variable names in 2000 Wave 2 by b and so on to j in 2008 Wave 10 For example gender of the main respondent is aSEX for Wave 1 bSEX for Wave 2 to jSEX in Wave 10 Serial numbers The unique household serial number SGERIALNO is assigned to each household in the family level files In the child level file each child has a household serial number SERIALNO and each child within a family has an identifier wPERSNO based on their position in the household grid These two variables can be used to create a unique respondent identifier which should be used to create merged files see matching files Interview participation alNT bINT and so on are wave specific indicators of whether a respondent has been interviewed in each wave year 1999 and 2000 respectively in this case Variable order Variables in the datasets are presented approximately in the order of the questionnaire by block The order of variables within each block is respondent variables in approximate questionnaire order followed by partner variables in the same order and lastly derived variables which are given in alphabetical order Edit imputation and extreme value flags are found next to the relevant variable Part
44. so includes information on the structure and content of interviews questionnaires and self completion questionnaires Chapter 5 presents information directly related to the datasets such as levels of data missing values variable naming and derived variables included with the dataset Chapter 6 provides a brief overview and practical advice on the cross sectional and longitudinal weights included on the dataset Chapter 7 contains various tips on how to use the data this should prove useful to the novice non expert user Chapter 8 includes a full list of publications and hyperlinks to FACS websites In addition to the user guide the FACS documentation includes a number of related documents that the reader is directed to throughout this guide These additional documents are Variable guide 2001 2008 This lists the SPSS variable names and labels under the main topics of the FACS questionnaire The guide shows the year in which each variable is present in the dataset Income variables guide This shows the structure of the income variables in a tree diagram how the components of household income relate to each other and how they feed into total household income Questionnaires 1999 2008 The full versions of the main respondent questionnaire and child self completion questionnaires are provided Derived Variables FACS datasets are accompanied by a core set of derived variables DVs derived to enable easier meaningful analy
45. t No 161 London Corporate Document Services McKay S 2004 Lone Parents in London Quantitative Analysis of Differences in Paid Work DWP In House Report No 136 Rafferty A 2003 The Characteristics of Lone and Coupled Mothers working Fewer than 16 Hours per week DWP In House Report No 125 Stephenson A 2001 Work and Welfare Attitudes Experiences and Behaviour of Nineteen Low Income Families DWP In House Report No 76 Willitts M and Swales K 2003 Characteristics of Large Families DWP ln House Report No 118 37 The above reports in the DWP Research Report Series can also be accessed online at http Awww dwp qov uk asd asd5 rrs index asp DWP Working Papers Brewer M and Shaw J 2004 Families and children Strategic Analysis Programme FACSAP Childcare use and mothers employment A review of British data sources DWP Working Paper No 16 Calandrino M 2003 Low Income and Deprivation in British Families DWP Working Paper No 10 Dorsett R and Kasparova D 2004 Families and children Strategic Analysis Programme FACSAP Low moderate income couples and the labour market DWP Working Paper No 15 Kasparova D 2006 Clusters of couples with children An exploration of their profile and transitions into work DWP Working Paper No 33 McKay S and Collard S 2004 Developing Deprivation Questions for the Family Resources Survey DWP Working Paper No 13 Phillips M Woodward D Colli
46. t the questionnaires Where a check was triggered the interviewer often opened and recorded a note explaining the respondent s situation These notes are recorded alongside the data and can be inspected by the research team The edit process involves the recoding of questions that include an other category Sometimes this generates a new response at other times the response suggests that one of the original categories would have been appropriate Values from these variables are backcoded into the original variables Imputation Missing data can arise because respondents either did not know the required answer Don t knows or were not prepared to provide an answer Refusals This is sometimes known as item non response Most imputation in FACS is based on replacing missing values with the median of the non missing values as detailed below In many instances the median among particular sub groups is used where sub group membership is known e g payment frequency is often known for earnings even if the value is not In general when imputations have been made the original variable has been overwritten but a flag variable has been created to inform the analyst how many cases have been affected by the imputation The flag variable contains the value that has been replaced following BHPS practice typically 998 in the case of a refusal and 999 for respondents who don t know the answer Imputation flags are suffi
47. tion A further concern with the FACS panel is the ageing out of the sample As an eligibility criteria of FACS is that a family must have a dependent child to participate it is expected The rescreen sample contains those families who were ineligible to participate in FACS in waves 1 and 2 likely because the family income was too high who became eligible to participate when the sample was redrawn in wave 3 7 Defined as a child in the household who is under 16 or 18 and under and in full time education 11 that a certain number of families will become ineligible to participate as their children grow older or leave the household There have been similar attrition rates amongst all seven of the yearly panel cohorts in FACS In general there is a drop off in respondents of between a quarter and one fifth after the first year and roughly ten per cent in years thereafter On average in FACS 78 per cent of any yearly cohort is interviewed again in the year following the first interview 71 per cent of cases are still in the study after two years 64 per cent after three years 56 per cent after four years about half 50 per cent remain after the sixth wave of interviews while to fifths 40 per cent remain after the eighth wave Table 3 12 Table 3 Longitudinal response as a percentage size of the original cohort and of each waves interviewed sample size Year of first interview FACS vo As vo As i to As
48. ts were men 153 cases the majority of whom 75 per cent were lone fathers For ease of interpretation the term mother is used to refer to the main respondent in this report This term is used to relate to the mother figure in However the nature of the questions are not sufficient to use this data to identify all instances where grandparents or other relatives are the primary carer for children 17 the family and therefore refers to lone fathers in lone parent families headed by a male unless otherwise specifically stated The partner respondent The partner respondent is the main respondent s resident husband or wife or cohabitee of the opposite sex Same sex couples are treated as two separate families because they are treated as two benefit units in the social security system The children would be allocated to the adult deemed to have main responsibility for them In 2007 and 2008 there was no interview with the partner and all partner data is from the proxy interview with the main respondent Dependent children A dependent child is defined as any resident child aged 16 years or under or aged 17 or 18 and in full time education The definition of in full time education used in FACS is made with reference to the end of the academic year ending the first Tuesday in September Thus a 17 or 18 year old child who had been in full time education during the current academic year was counted as being
49. ufacturing 4 Electricity gas and water supply 5 Construction 6 Retail hotels and catering 7 Transport and communication 8 Banking finance insurance business services amp leasing 9 other services incl health education and other public admin The variables to use in order to derive SIC codes 2001 from onwards are wsIC92 wSIC93 wSIC94 The variables to use in order to derive SOC codes are wXSC2000 wXSC2001 wXSC2002 The important difference in handling SIC and SOC data is that SOC data may be readily collapsed to the above schema but SIC may not So if the SOC code starts with a 1 whether a 2 3 or 4 digit number then it belongs in SOC major category 1 However SIC codes do not work like this and need to be reclassified using knowledge of the SIC 1992 hierarchy 32 For example manufacturing coded into division 3 below takes in codes 15xx 37xx The categorized variables are provided in the datasets wSICR9 respondent SIC groups wSICP9 partner SIC group main proxy combined wSOCR9 respondent SOC groups wSOCP9 partner SOC group direct proxy combined Childcare use A known problem exists whereby in 2000 not all respondents were asked about childcare during their interview Some cases were completed by a later follow up however there is still missing data on childcare for this wave year The variable bVERSION may provide a useful check in this regard as it rec
50. xed by X For example wLASTPYX is the imputation flag for wWLASTPAY last pay weekly The method of imputation the values that have been imputed and the number of imputed cases for FACS 1999 2008 are given in the relevant technical report 8 This section draws upon an earlier imputation note written by Steve McKay Many thanks for his permission to update that note for this report 14 The most common imputations related to housing costs savings income and debts The imputations included in the FACS series are pragmatic and facilitate analysis but users may wish to consider other approaches in particular settings Further data checks Some more complex checks require a separate in office editing and coding process using a modified version of the CAPI program The majority of these checks were consistency checks where responses in different parts of the questionnaire were unlikely to occur or were not logically possible according to some pre defined rule For example if a respondent was receiving a very high amount of benefits say 1 000 per week but the high amount was the result of interviewer error whereby an extra zero was accidentally added to the amount Where the editor was notified of such a problem he she was instructed to look for an interviewer note to help with its resolution In this case he she would correct the entry to 100 a week If this would not provide the solution to the check editors would
Download Pdf Manuals
Related Search