Census.ac.uk SARs 2011
Mar 28, 2015
Censu
s.ac.
uk
SARs 2011
Censu
s.ac.
uk
Where we arePhase 1: SARs user meeting 12 November 2007• consultation survey with users/non-users in 2007.
Phase 2: SARs user meeting 23 April 2009– Business Case by Keith Dugmore on the demand for the
2011 Census.– We proposed the 5 files/first document outlining the
principles underlying the strategy for 2011 census microdata presented to the ONS
Phase 3: SARs user meeting 28 Jan 2011• Produced file specifications and made available online • Consultation October 2010-Jan 2011. • Summary of consultations now online• Final consultation and feedback today!
Censu
s.ac.
uk
Key Feature of the 2011 SARs
• Coverage• Size• Ethnicity and religion• Geography• Multivariate• Household structure• Comparability with other census products• Easy access
Censu
s.ac.
uk
Importance of features 2011
Importance of file features 2011
0102030405060708090
100
feature
coun
t Not very importantFairly importantVery important
Censu
s.ac.
uk
Issues around data access
0
10
20
30
40
50
60
70
Per
cen
t
routine
exceptional
no
work in
instit-ution
institut-ion safe setting
analy-ses
checked
month long
applic- ation
travel away
Microdata research facilities survey 2009n(u) = 197Microdata users only
• Limited detail in EUL files• Institutions have
appropriate facilities• Biggest obstacle is long
waits• Travelling and having
analyses checked also unpopular
• Past CAM VML use is low• CAM users had mixed
experiences
Censu
s.ac.
uk
Recommendation of the Business Case
• Two popular files: for quick queries, education, and to attract new users
• Two professional files: offering more detail, but requiring more security
• Microdata Laboratory Facilities: offering considerable detail, but access only under strictest security con
• Production file: 15% household file – other files derived from ‘mothership’
Censu
s.ac.
uk
Access for 2011 SAR data
• Legal protection for personal data• Statistics Act establishes “Approved
Research Status”• High level spec assumed 3 levels of
security• ONS policy distinguishes 4 levels of
data (impact level)– Public use (0)– Licensed (1)– Contracted(2)– Identified (3)
• Secure Data Service is key for Approved Research Files
Balancing access against detail
Censu
s.ac.
uk
Statistical Disclosure and microdata policy
Level of SDC on the source data (protection via record swapping?)
Extra protection from sampling, imputation (item and non-response), churning (with time), , variable coding, large households?
SARs disclosure risk depends on other outputs (need to be considered in tandem)
Outputs policy in each country Different levels of SDC depending on access Special licences phased out? Secure Data Service (SDS) Virtual Microdata Lab (VML)
Censu
s.ac.
uk
2011 SpecificationsFile Sample
typeGeography Similar to Impact
levelAccess
1a General Regional File
3% indivi.
Region 2001 Individual Licensed SAR
1 licensced
EUL, available from CCSR
1b Local Authority File (EUL)
5% indivi.
Region 2001 Small Area Microdata file
1 EUL, available from CCSR
2a Individual File
3% indivi.
LAD/ UA/Metropolitan District/London Borough/ Parliamentary Constituencies (NI)
1991 Individual SARs, but with enhanced features
2contracted
Approved Researcher
SDS?
2b Household File
1% hierar-chal file
None 2001 Special License Household file.
2 Approved Researcher
SDS?
3. Controlled Access Microdata file.
Approx 15% HH
LAD/ UA/Metropolitan District/London Borough/ Parliamentary Constituencies (NI)
Facility to match small area indicators?
2001 CAMS files.
2 Approved Researcher
VML or SDS?
Censu
s.ac.
uk
Summary of the consultation on the variables for 2011• The specifications for the 5 files became available on the
SARs web page in Sep 2010:– Expert consultation in Sep. and open consultation in Dec 2010.– Most comments made by invited expert users, but also some
substantial comments by other users. – A total of 30 users commented on the variable specifications
for the different files
• Types of comments and actions required by the SARs team:– Comments that required simple modifications- some of which
have been adopted– Comments that needed further consultation and deliberation or
depend on sdc trade-offs between access/detail/other variables (some of these may be discussed here)
– Comments which are affected by circumstances beyond our control (e.g relating to the structure of the census questions)
Censu
s.ac.
uk
EUL LA file• COB — Country of birth (6 cat.)• AGE — Age of individual (0-89 singly, 90+ banded)• MARSTAT - Marital status (4 cat.)• DPCHUK11 - Household dependent children (not available)• BUSINESS Industry of employment (27 cat.)• ISCO — International Standard Classification of Occupations
(9 cat)• MAINLANG — Main language (2 cat)• ETHNICITY(NI) (2 cat.)• ETHNICITY (E&W) (18 cat)
Censu
s.ac.
uk
EUL Individual File
• COB — Country of birth: 16 cat. • DPCHUK11 Household dependent children (19
cat)• BUSINESS Industry of employment (17 cat.)• DTWPEWNI11 — Distance travelled to work (10
cat)• YRARR — Year of arrival in UK (14 cat.)• QUALVEWN Level of highest qualifications (6 cat.)
Censu
s.ac.
uk
Approved Individual File
• COB — Country of birth (49 cat)• DTWPEWNI11 — Distance travelled to work (0.1-
9999.9 km in 0.1 km units)• YRARR — Year of arrival in UK (Banded in 4 cat.
Until 1970, 1980+ in single years)• IDENTITY — National identity (49 cat)• ETHNICITY (E&W) (18 cat)• CITIZENSHIP — Passports held (7 cat)• MAINLANG — Main language (16 cat)• RELIGION(E&W) Religion (7 cat)
Censu
s.ac.
uk
Approved Household
• LARPUK11 Living arrangements (11 cat)• BUSINESS Industry of employment (98 cat)• ISCO — International Standard Classification of
Occupations (Full classification)• MOVEFROM — Address 1 year ago. (Banded
distance, 18 cat.)• ETHNICITY (E&W) (18 cat)
Censu
s.ac.
uk
CAMS
• BUSINESS Industry of employment (98 cat.)• ISCO — International Standard Classification of
Occupations (Full classification)• LAWRK —Local Authority District/ Unitary
Authorities /MD /London Borough/ Parliamentary Constituencies (NI) of Workplace (not available on EUL files)
• COB — Country of birth: Full country classifications