What is a Cohort Study?

Jane Elliott

Director of the Centre for Longitudinal Studies

July 2012

Sub-brand to go here

CLS is an ESRC Resource Centre based at the Institute of Education

ObjectivesIntroduction to the content and design of the British Birth Cohort

studies at CLS

Further examples of cohort studies

The distinction between a cohort study and a panel study

Examples of recent research using cohort data

Examples of how cohort studies have had an impact on policy

How to find out more/access the data

2

British Birth Cohort Studies

1946: MRC National Survey of Health & Development

1958: National Child Development Study

1970: 1970 British Birth Cohort Study

MCS: Millennium Cohort Study — first national birth cohort study for 30 years (2000-1)

Plans for a 2012/2013 cohort study

4

A Roman Cohort?

A cohort ....individuals moving or progressing as a unit

1958 Birth Cohort Study

Sample of over 17,000 infants born in 1958 (perinatal mortality study)

Sample followed at ages 7, 11, 16, 23, 33, 42, 46, 50 (biomedical at age 44)

Multipurpose study: family life; education; employment; skills; housing; health; finances; citizenship

Approximately 10,000 individuals are still participating

Mainly quantitative – highly structured interviews, but qualitative interviews with a subsample of cohort members at age 50

Core funded by ESRC with data collected every four years (five years after age 50)

PMS NCDS1 NCDS2 NCDS3 NCDS4 NCDS5 NCDS6 Biomedical NCDS7 NCDS8(1958) (1965) (1969) (1974) (1981) (1991) (2000) (2002-3) (2004-5) (2008-9)Birth 7 11 16 23 33 42 44-45 46 50

17,733a 16,883 16,835 16,915 16,457 15,600 15,145 12,037 11,739 12,316

Mother — Parents — Parents — Parents

School — School — School

Tests — Tests — Tests Tests

Medical — Medical — Medical — Medical

Subject — Subject — Subject — Subject — Subject — Subject — Subject — Subject — Subject

Census — Census

Spouse/ Partner Biometric measures

Consents to linkage

MothercBlood samples

Children Saliva sample

17,415b 15,425 15,337 14,647 12,537 11,407 11,419 9,349 9,534 9,793

Notesa: Target sample - Excludes emigrants, refusals & deaths. Includes immigrants at NCDS1-3.b: Achieved sample - At least on survey instrument partially completedc: Mother - Could be Cohort Member or spouse/partner

NCDS Follow-ups and sources of information 1958-2010Original sample: all living in GB born in one week in 1958

Biographical interview

Hypothetical life history

x

Born

1958

1st Child 1984 2nd Child

1987Age 7 Age 11 1991 2000

Age 42

2004

Age 46

Age 16

Age 23

1981

Age 33

Gets married

Parents’ social class

Parental interest in school work

Free school meals

Mother smoking

Parental divorce

Maths and reading tests

Teachers’ assessment of child’s behaviour

Exam results

Job 1 Job 2 Job 3

Voting behaviour

Psychological well being

Working hours preferences

Savings

Domestic division of labour

Union membership

Training and skills

Aspirations

Structured interview at age 50Household composition, marital status etc

Housing

Relationships & domestic division of labour

Children and parents

Family income

Employment and Partner’s employment

Pensions & attitudes to retirement

Qualifications, training and skills

Health (including menopause)

Alcohol consumption and smoking behaviour

Memory & Concentration

Voting behaviour, social participation & social support

Well-being

Consent to record linkage

Overview of the 1970 cohort study

• BCS70 is a multi-disciplinary longitudinal study following the lives of over 17,000 individuals born in Great Britain in one week in 1970

• There have so far been 8 sweeps of data collection: (Birth, 5, 10, 16, 26, 30, 34 and 38)

• The 9th sweep will take place in 2012 when study members will be aged 42: 75 minute face to face interview Additional self-completion questionnaire (paper or online)

• Funded by ESRC

Oversamples for Scotland, Wales and Northern Ireland, areas with high child poverty and, in England, areas with higher ethnic minority populations

Over 19,000 children born in the UK between September 2000 and January 2002

Follows children throughout their lives

Fifth survey, when children are in their last year of primary school

Funded by ESRC and UK government departments

One of five British Birth Cohort Studies

Overview of the Millennium Cohort Study

What have we done before? And what’s new?

Narrative elements of cohort studies

Allow us to trace lives through time & understand how childhood circumstances may impact on adult outcomes

Potentially allow for the construction of individual case studies based on detailed information collected over the years (while preserving confidentiality)

Allow for a focus on the historical context which has helped shape individual experiences

Comparisons between cohorts can enable the development of a narrative about social change

1958 and 1970 birth cohorts

0

10

20

30

40

50

60

1950 1960 1970 1980 1990 2000 2010 2020

Year

Ag

e

NCDS

BCS70

Cohort Comparisons

7

11

16

23

33

46

50

42

5

10

34

30

26

16

38

Life cycle effects

Source: Exploring Data (C. Marsh 1988) Figure 6.1 Unemployment as a problem in Britain: actual and perceivedSource: unemployed claimant count: Employment Gazette Decembet 1982 and May 1986. Percentage naming unemployment as most or second most urgent problem facing the country: Gallup Political Index monthly.

Proportion of women in paid employment, by age and cohort

Source: Jenny Neuburger - Paper presented at CLS June 2008

What is the difference between a cohort study and a (household) panel study?

Cohort Study1. All individuals in the sample are the

same age

2. Data collection focussed on specific life stage

3. Large sample of the same age facilitates modelling of specific life transitions

4. Focus on individual trajectories (limited info about other hhold members)

5. Data collection may be less frequent (e.g. British Birth Cohort Studies)

6. Objective health measures in childhood – resource for biomedical research

Panel Study1. Sample of adults representative of

whole age range

2. Questions need to be applicable to all

3. Sample spanning different ages allows for description of whole population

4. Focus on household dynamics and transfers

5. Data collection every year or perhaps every two years

6. Perhaps less likely to have a focus on health/bio-medical data

16

Qualitative Resources

•Children’s Essays

•Biographical Interviews at age 50

•Open ended question at age 50... ‘Imagine you are 60’

17

Social participation project: Qualitative Sub-study of NCDS 2008

• ESRC funded a qualitative component of the 2008 sweep of the 1958 cohort study

• Initial aim to interview 180 cohort members in 3 separate areas of Britain

• 170 interviews carried out in England and Scotland and a further 50 in Wales

• Aim to provide archived qualitative data for secondary analysis, together with the existing quantitative datasets

• Joint project with CRESC at Manchester: Jane Elliott; Andy Miles; Sam Parsons; Mike Savage

• Funded by the ESRC Research Resources Board

Structure of the interview and topic guide• Interview in six sections:

o Neighbourhood and belongingo Social participation and leisure activitieso Friendshipso Life story and trajectorieso Identitieso Experience of the NCDSo Visual exercises also included – personal community maps and life

trajectories

• Aim for an average of ninety minute interviews• Interviews digitally recorded and transcribed verbatim• Interviewers also wrote a brief summary of the interview• Interviews to be anonymised and deposited at the UK data archive

Age 50 self-completion questionnaire

At age 50, in 2008 NCDS Cohort members completed a 16-page paper self-completion questionnaire, including questions about health and well being and a personality inventory.

The final question stated:

Imagine that you are now 60 years old...please write a few lines about the life you are leading (your interests, your home life, your health and well-being and any work you may be doing).

Mean length of 7383 responses: 57 words

All of the 7383 responses have been transcribed and are being documented and prepared for deposit at the data archive.

Basic Skills – a major policy concern in Britain

22

The value of basic skills in the British labour market: Vignoles, De Coulon and Marcenaro-Guierrez (Oxford Economic papers 2011) Aim to evaluate the labour market value of literacy and numeracy

Draw on literacy and numeracy tests carried out with the BCS70 cohort at age 34 in 2004.

Make use of test score information collected during childhood and also information on qualifications and employment history in order to isolate the impact of basic skills on wages

Cross cohort analysis carried out to assess whether the wage return to skills has changed over time.

Richness of data from the cohort studies allows proper control for a wide range of observable characteristics.

23

24

25

26

27

The value of basic skills? Vignoles, De Coulon and Marcenaro-Guierrez (Oxford Economic papers 2011)

Models demonstrated that literacy and numeracy skills have a significant relationship with earnings even for individuals with similar levels of education

The effects appear to be very similar for both men and women

One standard deviation difference in skill levels is associated with approx 15% increase in earnings

Cross-cohort comparisons suggest that the value of basic skills has remained stable over time – this implies that the increase in supply of skills has been matched by an increase in demand for skilled workers

28

Poverty, family resources and children's early educational attainment: the mediating role of parenting (Kiernan and Mensah, 2011, British Ed Research Journal)

• Uses longitudinal data from first three waves of MCS (N=5462)• Focus on both episodic and persistent poverty• Outcome measure is based on the Foundation Stage Profile collected from

teachers when children were aged 4-5 in the first year of primary school• Equivalised family income was used to measure poverty• Family resources index was also constructed: income poverty, mother’s

education, family employment, housing tenure, quality of the local area, mother’s age at birth of her first child, family structure, number of children in the household, child’s birth order, child’s ethnic origin and the language spoken in the home

• Parenting measure constructed using reports of activities with child and interviewer observation

29

Poverty, family resources and children's early educational attainment (results summary)Children from poor families and those with low levels of family resources

are doing worse in their first year of school

Poverty matters but persistent poverty is even more detrimental

The parenting index was also a very important factor predicting children’s early educational attainment

Positive parenting matters regardless of the levels of resources in the family

A decompositional analysis suggested that about one-half of the effects of child poverty and 40% of resource disadvantage may be accounted for by the quality of parenting the child has received in early childhood

30

How do I access the data?Data is available via the Economic and Social Data Service based at the

University of Essex

Data is free for non-commercial use (e.g. teaching and research)

Following a simple registration procedure data can be downloaded directly from the Economic and Social Data Service website

Information about what questions have been asked is provided via the CLS website but we can also answer individual enquiries about specific topics.

Plans to make more potentially disclosive data available via the Secure Data Service at the University of Essex:

http://securedata.data-archive.ac.uk/home

CLS Birth Cohort Studies: Web ResourcesResources available via CLS website:

(www.cls.ioe.ac.uk/Cohort/mainncds.htm)– Searchable bibliography (over 2000 articles, chapters etc)– Briefings (summaries of recent findings)– Annotated Questionnaires and CAPI Documentation– Technical Reports – e.g. on sampling, instrument development and

fieldwork of MCS– Data Dictionaries – Events and workshops

Policy relevance of 1958 and 1970 cohort studiesThe health impact of smoking in pregnancy

Child poverty

Declining intergenerational mobility

Antecedents and consequences of disability

Health continuities over the lifecourse and health inequalities

Determinants of crime and anti-social behaviour

Social and economic returns to education and training

Access and barriers to higher education

Improving adult basic skills

Women’s opportunities in employment

Maternal employment and child outcomes

Conclusions: research questions best addressed by birth cohort data

Long term outcomes of experiences and decisions in early life

Medium and short-term outcomes & links between different life domains (e.g. health and employment)

Descriptions of individual trajectories – careers, relationships, fertility, poverty and disadvantage

The links between social change and the changing experiences of different cohorts

Intergenerational transmission of advantage and disadvantage and the processes involved

Websitewww.cls.ioe.ac.uk

Please register for regular updates

Follow us on twitter at:

www.twitter.com/Clscohorts

Overview of the Cohort Resources Facility

36

37

Leadership team

WP6: Data linkage - geography

WP5: Data linkage administrative data

WP7: Data linkage – health data

1946 cohort

1958 cohort

1970 cohort

ALSPAC

MCSUnderstanding

Society

SWS

HCS

New BCS

Cohorts’ CollegeCohorts’

repository

Metadata

Uniform Search Platform

Training and capacity building

Impact

BCS70 Sample size (Previous sweeps) - additional information

0 (1970)

5 (1975)

10 (1980)

16 (1986)

26 (1996)

30 (2000)

34 (2004)

38 (2008)

Total 16,571 13,071 14,874 11,621 9,003 11,621 9,665 8,874Postal Survey

Telephone Survey

Achieved sample size in previous sweeps

N % Eligible

Productive 8874 75.6

Non-contact 1949 16.6

Refusal 712 6.1

Other unproductive 198 1.7

Ineligible (Died / Emigrated) 110 -

Total Issued Sample 11,843

Age 38 Telephone Survey Response

A different cohort design – the Hertfordshire Cohort Study

39

Hertfordshire Cohort Study: structure

Live births in Hertfordshire recorded in Health Visitor’s ledgers39,764

Traced and living in Hertfordshire8,650

Home interviews3,225

Clinical outcome questionnaire2,225

Dead215

Musculo-skeletal follow-up642

Clinic visits2,997

Hertfordshire Cohort Study: measurements

Historical records: Birth weight, weight at one yearInfant feeding/infections

Questionnaire: General healthHRQoL: SF36Diet: FFQ and 24 hr diaryPhysical activity

Clinic visit: CVD: blood pressure, ECGCOPD: spirometry

Type 2 DM: anthropometry, OGTT Osteoporosis: DXA

OA: Hand & knee x-raySarcopenia:strength, anthropometry,

pQCTFrailty: Fried score, Rockwood indexWellbeing: WEMWB Scale

Venous blood: Glucose, insulin, bone turnover, DNA

Syddall et al Int J Epidemiol 2005

Hertfordshire Cohort Study: geography

Hertford

Radlett

Royston

Welwyn Garden City

Southampton

Baldock

Hertfordshire Cohort Study: history