310 likes | 322 Views
This seminar delves into analyzing household life-courses, transitions, and intra-relations using longitudinal data from studies such as the BHPS in the UK and internationally. Explore data quality improvements, household panel surveys, and longitudinal questions.
E N D
Household Studies Paul LambertStirling UniversityPrepared for “Longitudinal Data Analysis for Social Science Researchers: Introductory Seminar”, Royal Statistical Society, 28th April 2006
Quantitative Longitudinal Household Studies • Repeated cross-sectional household surveys • Longitudinal surveys with household information • Household panel surveys • The BHPS in international context • The BHPS in the UK April 2006: LDA
Longitudinal Households? • .. device to talk about the BHPS • Improving data quality • Reliability: Household sharers’ records • Context: Household sharers’ data • Longitudinal questions • Household life-courses • Household transitions • Intra-relations: similarity and dependence April 2006: LDA
1) Repeated cross-sectional household surveys for example.. • UK Census • Labour Force Survey • General Household Survey • Family Resources Survey • ??Annual population survey April 2006: LDA
2) Longitudinal Surveys with Household Information • Cohort studies (see later) • Birth Cohort Studies • Youth Cohort Study • Census Longitudinal Studies (see later) • Individual panel studies • ELSA • Re-contact studies • Individual level repeated cross-sections • Social attitudes surveys • ISSP / WVS / ESS • British Election Studies April 2006: LDA
3) Household Panel Surveys • The BHPS • The BHPS in international context • The BHPS in the UK April 2006: LDA
The British Household Panel Survey 1991-2006 • Panel study of 5k households re-contacted annually since 1991 • Major UK research investment For lots more introductions, see: http://www.longitudinal.stir.ac.uk/ April 2006: LDA
The ‘Essex’ BHPS • ISER - Institute for Social and Economic Research • ULSC - UK Longitudinal Studies Centre • Design, coordinate, release, analyse and promote the BHPS • Data supplied by UK Data Archive at University of Essex • Online documentation and support: http://iserwww.essex.ac.uk/ulsc/bhps/doc/ April 2006: LDA
Annual survey since 1991 • Sample re-interviewed once a year • Each new panel is a ‘wave’ • Interviews start each September • Datasets updated and re-released annually • Government funding to at least 2009 April 2006: LDA
BHPS data file structures April 2006: LDA
You’ll most likely use.. • Adult individual interviews • All adults within household contribute and individual record • Youth records • All 11-14’s within houshold • Combined life-history files • Oriented around event history analyses (durations) April 2006: LDA
Sampling design • W1 (1991): Stratified random sample of 5,500 households • 14,000 ‘OSM’ household members • Later waves: trace all OSM’s; their descendants; and their household sharers (TSM’s) NB: longitudinal trace of individuals and their surrounding household, but not of ‘longitudinal households’ April 2006: LDA
W7-11 -> ECHP supplement (low incomes) W9- -> Scottish and Welsh boosts W11- -> Northern Irish boosts Future: possible minority group boosts? These are important!! affect representativeness use of weights is complicated catches every user out at least once… Extension samples April 2006: LDA
BHPS Unbalanced panel & Data Management: Below data may have come from 6 different BHPS source files
The household structure of the BHPS • All adults within a household are interviewed • Clustering analysis issues • Person groups? • All persons within a household are ennumerated • Children records • Rising 16’s • Siblings and migration • BHPS Household analysis possibilities are exciting but complex.. April 2006: LDA
The BHPS in International Context • Part of: • ECHP (1997-2001) • CHER (1991-2000) • PACO (1991-1998) • CNEF (1991->) • EU-SILC (2003 onwards: under-discussion) • Numerous stand-alone comparative projects • Source project : US PSID • Use of comparable questionnaire design April 2006: LDA
Cross-national comparisons • Focussed studies • McGinnity, F. (2002) “The Labour-Force Participation of the Wives of Unemployed Men: Comparing Britian and West Germany” European Sociological Review, 18(4)473-488 • Benefits system influences wives participation in Britian, not Germany • Broader comparisons • Robson, K. and Berthoud, R. (2003) “Teenage Motherhood in Europe: A Multi-Country Analysis of Socioeconomic Outcomes” European Sociological Review, 19(5)451-466 • Substantial variations in economic circumstances of teenage mothers, and their family structures, across Europe April 2006: LDA
The BHPS in the UK • Major ESRC investment • 5500 households nationally • Scottish, Welsh and Northern Irish boost samples 1999 / 2001 April 2006: LDA
Assets of the BHPS 1) Large scale panel dataset! • Answer questions on processes, transitions, state dependence • Offers longitudinal controls • Extensive methodological resources, support, validation 2) Household information • Full information of family sharers • Complex but tractable handling April 2006: LDA
Assets of the BHPS 3) The wYOUTH records • Unique datasource on 11-15yrs • Can link to family / trace to adulthood 4) Occupational information • Detail on own jobs, histories, spells • Detail on family, parental, friends jobs April 2006: LDA
Assets of the BHPS 5) High quality income information • Range of sources • Imputation procedures / validation work 6) Subpopulaton analyses • Plausible to identify and analyse distinct groupings, eg home nations • But not others, eg ethnicity April 2006: LDA
Drawbacks with the BHPS 1) Complexity of the data records • Puts off potential users • Forces advanced users to specialise 2) Short term panel coverage • Doesn’t yet span long enough spells April 2006: LDA
Drawbacks with the BHPS 3) Dropout and item non-response • A little of the first • A lot of the second 4) Interviewers & Panel conditioning • Panel conditioning very likely? April 2006: LDA
Drawbacks with the BHPS 5) Regional sampling bases • Endogeneity to labour market? • Imposition on generalisations, eg more from Dundee than Glasgow . 6) Complex clustering • BHPS individual level response = Ytijkl April 2006: LDA
Three modes of analysis 1) (Repeated) Cross-sectional 2) Panel 3) Life history April 2006: LDA
Example: Descriptive analysis of panel data April 2006: LDA
SUMMARY • BHPS many complexities • But more potential uses • Eclectic resources • Greater use seems sensible April 2006: LDA