Review of Panel and Cohort Studies

1
Review of Panel and Cohort Studies

• Summary of conclusions and recommendations
• National Longitudinal Strategy Committee
• 17 January 2006

2
Review Team

• Jean Martin
• John Bynner
• Harvey Goldstein
• Graham Kalton
• Paul Boyle
• Vernon Gayle
• Andrea Piesse
• Sam Parsons

3
Aims of the Review

• Make scientific case for large scale longitudinal
  data resources
• Review ESRCs existing portfolio and develop
  future strategy
• BHPS
• 1958, 1970 and Millennium birth cohort studies
• Other studies
• New studies
• Assess balance of different types of longitudinal
  resources needed
• Consider methodological and other challenges and
  appropriate approaches

4
Background

• ESRC bid for longitudinal resources Large
  Science Facilities Bid
• UK Data Forum co-ordination between different
  funders
• Emphasis on strategic and long-term thinking

5
A National Framework

• UK Data Forum/NLSC roles
• Portfolio of longitudinal studies
• ESRC major stakeholder
• Multi-disciplinary, multi-level,
  intergenerational
• Long-term strategy
• Science-based rather than policy driven
• Oversight of
• Review/continuation of existing studies
• Enhancement to meet new needs
• Development of new studies

6
Sources of Evidence

• Interviews, meetings, telephone and
  correspondence with over 70 stakeholders/experts
• Two-day consultative conference involving 42
  experts
• Website and other sources of documentation
  details for 91 major longitudinal studies in the
  UK and overseas

7
The Report

• Background and method
• Case for investment
• Longitudinal research designs
• Portfolio of longitudinal resources
• Development
• Data quality methodology and analysis
• Building new capacity
• Access and disclosure
• Governance and funding
• Collaboration

8
Case for Investment National Challenges

• Ageing population
• Long-term effects of childhood experience in
  changing social and physical environment
• Changes in timing and duration of life course
  transitions
• Bio-techology revolution
• Shifts in family formation and structures
• Mobility, immigration and emigration
• Dynamics of diversity and inequality
• Effects of globalisation

9
Research Questions and Types of Design

• Long term developmental processes, e.g. obesity,
  ageing
• Birth cohorts
• Family dynamics and transfers, e. g. poverty and
  inequality
• Household panels
• Transitions, e.g. retirement
• Age cohorts
• Transient statuses e.g. immigration
• Short-term panels

10
Design Decisions

• Topic coverage
• interdisciplinary
• Population coverage
• general, subgroup, cohort/age group, area
• Sample design
• over-sampling, refreshment, clustering
• Data collection
• duration, frequency, topic rotation, mode
• Additions
• tests, diaries, qualitative, admin data, biomed
  data
• Harmonisation comparability

11
British Household Panel Study Limitations

• Small sample size
• Cant study small subgroups e.g. ethnic
  minorities, lone parents
• Small numbers of rare events e.g. household
  breakdown/formations
• Regional comparisons limited
• Aggregating data across years confounds period
  effects
• Recent immigrants excluded
• Attrition is a concern re bias and reduced sample
  size

12
Design Options

• Need for very much larger panel
• Completely separate new panel?
• Enhancement of current BHPS?
• Designed to allow comparison and combined
  analysis
• Review
• topic coverage
• frequency of data collection
• use of rotating modules/sub-samples
• Need task force to do detailed design work

13
Number of events per annum
14
Ethnic groups expected nos of adults
15
Birth Cohort StudiesLimitations

• Infrequent data collection
• Lack of area level data (except MCS)
• Limited info re other family members
• Attrition increasing problem
• No immigrants after age 16
• 30 year gap between 1970 cohort and MCS

16
Birth Cohort Studies

• Major gains from extending studies and analysing
  cohorts together
• Expand support for existing studies including
  intergenerational component
• Start new study around 2012
• Consider new design options
• Spread over several years?
• More than one starting age?
• Links to admin data
• More biomed data
• Review potential of ALSPAC/LSYPE to fill missing
  cohort gap

17
Birth cohort studies continuities and gaps
Age
60
55
50
NSHD
45
NCDS
40
BCS70
35
30
25
20
15
10
MCS
Child Data
5
0
1940
1950
1960
1970
1980
1990
2000
2010
Year
18
Other Longitudinal Studies

• Not necessarily ESRC led (cf ELSA)
• More specialised studies funded by main
  stakeholders
• Ethnic minorities
• Immigrants?
• Transition to parenthood?

19
Strategic Planning

• Develop major studies as frameworks for more
  specialised longitudinal studies
• Build in data comparability/harmonisation
• Design trade-offs
• Link in/collect other types of data
• Institutions
• Other admin
• Blood sampling
• Environmental measures

20
Major challenges

• Methodology,
• Attrition, missing data, measurement error, panel
  maintenance, development studies methodological
  panels
• Capacity building
• Training opportunities, training constituencies
  
• Data access and disclosure
• Sensitive data, disclosure risk

21
Governance and Funding

• Preservation of scientific integrity
• Long term funding base
• Steering and expert advice
• Collective memory

22
Collaboration

• Between longitudinal study teams
• Across scientific disciplines
• Between analysts
• Between academics and policy researchers
• Between UK countries
• Internationally
• Annual standing conference