Handling Migration and Handling Migration and Commuting Flow DataCommuting Flow Data
Day session at the ESRC Research Methods Festival at St Catherine’s College,
University of Oxford, 2 July 2008
Centre for Census Interaction Data Estimation and ResearchCentre for Census Interaction Data Estimation and Research
Programme: Programme: MMorning session
Data Issues and Estimation Methods
• Sources of interaction data in the UK• Web-based Interface to Census Interaction Data• Shortcomings of Census Interaction Data
• Estimating Spatially Consistent Interaction Flows• Using Interaction Data from the SARs and the LS• Measuring Ethnic Migration Using Commissioned
Data
Programme:Programme: Afternoon session
Analysis and Modelling Methods
• Migration and Socioeconomic Change• Commuting in Rural Areas• Modelling of Migration Flow Tables using Log
Linear Models
• Poisson Modelling of Migration• Spatial Interaction Modelling of Commuting to
School• Finding Clusters Among the 25 million
Commuters in a Billion Cell Matrix
Sources of Interaction Data Sources of Interaction Data in the UKin the UK
John Stillwell Centre of Interaction Data Estimation and Research
School of Geography, University of Leeds
Presentation at the session on ‘Handling Migration and Commuting Flow Data’ at the ESRC
Research Methods Festival, St Catherine’s College, University of Oxford, 2 July 2008
Aims
• Research methods are frequently determined by the type of migration and commuting interaction data that are (not) available (macro/micro data)
• Important for researchers to know what data sets exist and what are the characteristics of the data from different sources
• CIDER undertook an audit of interaction data (Dennett et al., 2007, Working Paper 07/05, School of Geography, University of Leeds) –
• This presentation is an overview of interaction data sources
PresentationPresentation
• Types of data source
• Definition and measurement
• Census interaction data sources
• Administrative interaction data sources
• Survey interaction data sources
• Conclusions
Interaction data sourcesInteraction data sources
• Censuses of Population – comprehensive and reliable migration and commuting data, particularly for flows within and between small areas
• Administrative records – collection of records arising from some transaction, registration, or record of service delivery
• Social surveys – samples of population allowing useful cross-classification at national (regional) level
Definition and measurementDefinition and measurement
• Migration and commuting variables defined in different ways in different sources
• How is a migration defined? A permanent change of usual residence? What about those moving between residences on a temporary basis: students, second home owners, etc
• How do we measure commuting? The journey from home to work? What about those who work in different places or those who travel from home to temporary accommodation for periods of work, etc
We tend to think of migration and commuting in the way they are derived from census questions
• These definitions and measures are particular to the census but there are different measures of migration: e.g. migrants, wholly moving households, moving groups, representative persons of moving groups
• More generally, an important conceptual difference is between migrants/transitions (census data) and migrations/moves/events (registration data)
Migrants versus moves: Lexis representationMigrants versus moves: Lexis representation
FBDE includes persons from the 1986-91 birth cohort who were all aged 5-9 in 1996 and 10-14 in 2001; these are counts of exist-survive migrants over the five year period.
ABDF includes those aged 10-14 during the period 1996-2001 (including half those from the 1986-91 birth cohort and half those from the previous cohort); these are counts of moves over the five year period
Main Census tablesMain Census tablesCounts of migration are available from the following tables:Key Statistics Table 24, Standard Tables 008 and 009 and Theme Table 033
But no tables on commuting
Origin-Origin-DestinationDestinationStatistics Statistics 2001 and 2001 and 19911991
SMS
SWS
STS
Tony Champion, Robin Flowerdew
Martin Frost, Mike Coombes
Adam DennettOliver Duke-Williams
Zhiqiang Feng
James Raymer
Migrants and commuters to Leeds from 2001 SMS and SWS
Daily in-commuters in 2001
In-migrants in 2000-01
Commissioned tablesCommissioned tables
• Customised output from the 2001 Census may be commissioned from ONS Customer Services when particular cross-tabulations are not available from the standard tables
• Commissioned tables incur charges to recover staff and material costs
• Once delivered and paid for by customer, each table listed on the ONS website and available to all users free
• All commissioned tables of 2001 data are subject to checks to ensure confidentiality - each table is subject to SCAM procedures and consequently inconsistencies will appear when checking totals with data from other census sources
Serena Hussain
Samples of Anonymised RecordsSamples of Anonymised Records• Microdata relating to individuals or households
1991: • Individual SAR (2% sample) • Household SAR (1% sample)
2001: • Individual SAR (Licensed) (3% sample)• Household SAR (Licensed) (1% sample)• Individual Controlled Access Microdata Sample (Individual
CAMS)• Household Controlled Access Microdata Sample
(Household CAMS) • Small Area Microdata (SAM) (5% sample)
• Spatial resolution is key issue with micro data: for migration analysis, the 2001 SAM has the advantage of a detailed destination geography, although the origins remain as GOR
Paul Norman
Longitudinal StudiesLongitudinal Studies• Sources that contain multiple observations of a population
of interest over a period of time
• ONS Longitudinal Study of England and Wales (LS) • Scottish Longitudinal Study (SLS) • Northern Ireland Longitudinal Study (NILS)
• These differ in a variety of ways including the length of the time period covered, the sampling fraction used, and the types of other data linked into the study
• New UK Household Longitudinal Study (UKLHS) - first wave of data collection in 2008 consisting of a wholly new sample of households (40,000), an ethnic minority boost sample, and a sample (up to 100%) drawn from the existing British Household Panel Survey (BHPS)
Paul Norman
NHSCR data
• NHS Central Register (NHSCR) at Southport records movements of patients between Health Authority (HAs) areas in England and Wales and the Census Office has developed systems for capturing the reporting of re-registrations between areas
• Entries in the NHSCR include the date of birth, sex, the codes of origin HA and destination HA
• NHSCR data are ‘movement’ data rather than ‘transition’
data • Time series available from mid-1970s but problems of
changing administrative units/boundaries – especially in late 1990s
Changing patterns Changing patterns of net migration as of net migration as shown by NHSCR shown by NHSCR datadata
1988-90
1980-82
Source: Stillwell (1994)Environment and Planning A
Patient Register dataPatient Register data
• NHSCR system in England and Wales only records movements between HAs - ONS has in the past used information from electoral registers and the census to apportion inflows and outflows between constituent local authorities (LADs)
• The inadequacy of the electoral registers in the estimation of sub-HA flows led ONS to investigate the patient registers held by every HA in England and Wales
• Registers contain the NHS number, gender, date of birth, date of acceptance at the HA and, importantly, the postcode of address, for each patient
• With postcode unit information being available, it is possible theoretically to create aggregate migration matrices for any level of geography
Patient Register dataPatient Register data
• Patient Register Data System (PRDS) contains an annual download from every patient register
• Comparing records in one year with those of the previous year by linking on NHS number enables identification of each person who changes their postcode
• Annual ‘migrants’ rather than ‘moves’ data • Range of tables available from mid-1998 including
flows between local authority districts by broad age group
• Scotland and Northern Ireland have independent systems
Comparison of PR Comparison of PR and Census and Census migration at district migration at district level, England and level, England and Wales, 2000-01Wales, 2000-01
-10000
-8000
-6000
-4000
-2000
0
2000
4000
6000
-6000 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000
PR
net
mig
rati
on
Census net migration
0
5000
10000
15000
20000
25000
30000
35000
0 5000 10000 15000 20000 25000 30000 35000
Ce
ns
us
in-m
igra
tio
n
PR in-migration
NET
IN
OUT
0
5000
10000
15000
20000
25000
30000
35000
40000
0 5000 10000 15000 20000 25000 30000 35000 40000
Cen
sus
ou
t-m
igra
tio
n
PR out-migration
r = 0.783
r = 0.985
r = 0.987
Changing patterns of net migration as shown Changing patterns of net migration as shown by patient registration databy patient registration data
2005-06 balances Changes in net balances2000/01-2005/06
PLASCPLASC• Whilst the 2001 STS in Scotland provided details of the daily travel to study for students and children, similar data are not produced for England and Wales or Northern Ireland
• Pupil Level Annual School Census (PLASC) collects data from each local education authority (LEA) in England and Wales on the location of pupils and the schools that they attend
PLASC dataPLASC data
• PLASC is the foundation of the National Pupil Database (NPD) and includes variables such as age, sex, ethnicity, free school meals and special education needs
• Linking of pupils from one year to the next using a ‘unique pupil number’ (UPN) means that a longitudinal profile of each pupil is available
• Potential for pupils to be tracked over time and their transitions through the education system to be identified, including their movements between schools and between different home addresses
• PLASC data are therefore a potential source of data on (i) commuting to school, (ii) ‘pupil mobility’ between schools and (iii) ‘child migration’ from one usual residence to another
Kirk Harland
Example of PLASC Example of PLASC data for Leeds used to data for Leeds used to measure residential measure residential migration and pupil migration and pupil mobilitymobility
Data source: Education Leeds
Pupils moving home by age
Hospital Episode Statistics (HES) Hospital Episode Statistics (HES)
• HES data include details of all patient admissions to NHS hospitals in England from 1989-90 onwards
• Each record holds around 100 personal, medical and administrative details of each patient admitted to hospital in England, including geographical information about the location of treatment and where the patient lived
• Around 12 million new records added to the dataset each year
• Requests for data in the form of database extracts or custom tabulations are made to the NHS Information Centre through their external data custodians, Northgate Information Solutions
• Potential source of information about commuting to hospital
HES dataHES data
Total number on in-patient visits (including repeats) made to Yeovil and Weston Hospitals from wards of England in 2000/01
SurveysSurveys• principal advantage is that they publish results
with relatively high frequency – often annually, but in some cases quarterly, allowing creation of data time series
• major drawback is the spatial resolution for published statistics - tends to be the GOR - so only very general patterns can be observed, despite the rich variety of other attributes that can be ascribed to the individual respondents
• the sample size of the survey is often relatively small, with implications for representativeness
ConclusionsConclusions
• Census products remain the most important data sources for migration and commuting flows but other sources, particularly administrative, have potential for greater use in ‘interaction’ research e.g. patient registers, PLASC and HES
• Questions to be asked in next census for England and Wales on 27 March 2011 not yet agreed but might generate some new interaction data – on migration or mobility of individual residents – and on origins of visitors
Questions from Test Census on 13 May 2007Questions from Test Census on 13 May 2007
http://www.statistics.gov.uk/census/2011census/CollectingtheInfo/questionnairedevelopment.asp
Beyond 2011Beyond 2011• 2011 is likely to be the last census in the UK• ONS has proposals for an integrated population statistics
system (IPSS) that combines census data at individual level into a single comprehensive statistics database with survey and administrative data and will underpin the country’s population and social statistics
• Includes a high quality address register containing information on properties and characteristics of individuals associated with these properties together with a population register, which will provide the basis for linkage with data from other sources
• If it happens, IPSS likely to be source of interaction data on a more regular basis - will be very important to ensure that data release is maximised without the effects of disclosure control becoming too detrimental