What is the ONS Longitudinal Study? (2)
The data in the LS come from complete census returns from 1971 onwards, linked with vital event data from various sources. The census data in the LS, which is available for both LS members and the people they live with, include the following:
- Age, sex, marital status
- Family, household or communal establishment type
- Housing, including tenure, rooms and amenities
- Country of birth, and (in 1971) parent's country of birth
- Ethnicity (1991 & 2001)
- Educational qualifications
- Economic activity
- Occupation and social class
- Migration and travel to work
- Long-standing illness (1991 & 2001) and self-rated health (2001)
- Religion (2001)
- Caregiving (2001)
- Marriage and fertility history (1971)
Copies of the census forms used to collect these data are available for download here.
The events data linked to LS members' census records mean that the data set also includes information on:
- Fertility (live and still births registered to women, and to men for the period 1971-78 and 1981 only)
- Infant deaths (deaths of infants born to sample mothers)
- Cancer registrations
- Mortality (deaths of sample members), including full information from registration forms, such as cause of death
- Widow(er)hoods (deaths of sample members' spouses)
As an example, consider this timeline which shows census and event data recorded for an hypothetical LS member.
An overview of the structure of the LS is available (PPT 42K), showing the different types and numbers of events linked into the study. A poster (in .pdf format) summarises the structure and offers examples of projects using the LS.
The list below shows the various research areas in which the LS has been used. Click on an item in the list to see a list of publications in that research area.
- Infant mortality
- Living arrangements
- Older people
You could also view a presentation: Academic research using the ONS Longitudinal Study by Emily Grundy, March 2007 (PowerPoint 765K).
One of the major strengths of the ONS Longitudinal Study is its size. It represents 1% of the population of England and Wales, with a sample of about 950,000, and covers over 30 years of census and event records. This combination of large sample size and long follow-up time makes it unique. For studies in certain areas, such as ethnicity, the LS is the only longitudinal study large enough to give statistically useful sample sizes. Other important advantages are that the data set includes information on co-residents of LS members as well as the members themselves, and information on people in communal establishments, who are excluded from most national surveys. Census coverage rates and the linkage and trace rates obtained in the LS compare very favourably with response rates achieved in surveys.
Its main limitation is the relatively restricted range of variables available. For example, there is no information on smoking or on income. It is not possible to go to the sample and gather information on extra variables, as is sometimes the case when dealing with surveys. However, it is possible to link information on the characteristics of areas that LS members live in (for example, deprivation score of the ward of residence). The LS is also restricted to England and Wales, though a Scottish LS exists and guidance is being developed for comparative research. A Northern Ireland Longitudinal Study has recently been launched.
For certain studies, another data set may be more suitable than the LS. Some suggestions are provided below:
- Longitudinal studies:
British Cohort Studies:
1946 (MRC National Survey of Health and Development): cohort of 5,362 births in 1946 followed up to the present day. Contains data on subjects' physical and mental health, including observational measures, diet and lifestyle, and family, employment and social circumstances.
1958 (National Child Development Study): cohort of 17,000 births in 1958 followed up at intervals (most recently 2008-9). Contains a wide range of medical/health data.
1970: cohort of 17,198 births in the UK in the week 5-11 April, 1970 followed up to 2012 so far. Contains medical data and data on physical, educational, social and economic development.
Millennium Cohort Study: follows a cohort of 19,000 children born in the UK during a twelve month period from June 2001. Fieldwork started in September 2003 in England and Wales, and December 2003 in Scotland and Northern Ireland.
British Household Panel Survey: cohort of 5,500 households and 10,300 individuals in Great Britain, followed up annually from 1991. Sample refreshed by births to sample members. Now expanded to become:
Understanding Society: a panel study of the socio-economic circumstances and attitudes of people in 40,000 households (including British Household Panel Survey sample members). First data collection took place in 2009.
English Longitudinal Study of Ageing: started in 2002 with a sample of 12,000 people aged 50 or over and their younger partners. Data collection every two years, sample refreshed at intervals.
More information on these, and other longitudinal surveys, is available from the Keeping Track website at the Institute for Social and Economic Research (ISER) at the University of Essex. ONS published a list of longitudinal sources in 1999 which is still worth consulting though somewhat out of date.
- Cross-sectional studies:
Samples of Anonymised Records (SARs): a 3% sample of census records at individual level, and a 1% sample at the household level, available separately for Great Britain and Northern Ireland from the 2001 Census. Similar files availble for the 1991 Census. There is also a 5% sample for 2001, with more detailed geography but less detail in other characteristics.
General Household Survey (briefly the General Lifestyle Survey): annual survey based on a sample of around 10,000 private households in Great Britain (around 18,000 adults). Started in 1971; from 2009 it became part of the new Integrated Household Survey and included a longitudinal element, but in 2011 it was announced that the General Lifestyle Survey would be discontinued.
Health Survey for England: annual surveys with a common core of questions and specific modules with a different focus for each year. Questions asked and sample populations vary from year to year depending on the focus.
Labour Force Survey: a quarterly sample survey of 60,000 private households in Great Britain (with an equivalent in Northern Ireland), followed up over 15 months. Questions asked on respondents personal circumstances and their labour market status. Now part of the Integrated Household Survey.