Towards a high quality 2011 Census - The 2011 Census Questionnaire

Download Report

Transcript Towards a high quality 2011 Census - The 2011 Census Questionnaire

Towards a high quality 2011 Census
The 2011 Census Questionnaire
Pete Benton
Deputy Director, Census Programme
Overview
• Quick look at
– Development history
– Questionnaire content
• How does it all come together to produce a
high quality population estimate?
• What about short term migrants?
Census provides statistics on:
• Population units:
– people and housing &
– key demographics (age, sex, marital status,
ethnicity)
• Population structures:
– households, families
• More detailed characteristics :
– eg religion, labour market status, industry,
qualifications, health/disability; etc
• Key requirement to ‘Get the Count Right’
Key lessons from 2001 Census
• Need to gather information on more than
just usual residents
– Include visitors
• Need more information to aid
understand of coverage
– Addresses/households
– People within households
Subsequent changes in society
• More people at more than one address
– Weekday residences for work
– Holiday / weekend homes
– Children of divorced parents
• International migration
– More ‘short term visitors to UK’ – resident or not?
• Plus familiar issues of students, armed
forces, prisons, hospitals, hotels, hostels
• Risk of undercount, overcount, or wrong location
• Need to count the right people, in the right place
– and be able to demonstrate this with confidence
Questionnaire research to date
• Formal 3 month consultation (May 2005)
• Over 2000 responses from 500 users
• Scoring of user requirements
• Initial topic proposals
• Further topic-specific consultation
• Roadshows
• Question testing
– Qualitative: Cognitive testing (6 waves over 3 years),
focus groups
– Quantitative: 2007 test, omnibus survey, postal tests
Summary of user requirements
• Most 2001 topics
• Many new topics, including:
- income (not included)
- language
- second residences
- sexual identity (not included)
- national identity
• More than 3 pages of questions!
• Difficult trade-offs to be made
• Additional £22m funding for 4th page of
questions per person obtained
New topics - population
• Population units and structures
–
–
–
–
National identity
Citizenship
Visitors
Year / month of entry into UK
• Intended length of stay??
– Second residence address and purpose
• Population characteristics
– Language
New topics - housing
• Housing characteristics
– Number of bedrooms
– Type of central heating (just ‘yes/no’ in 2001)
Removed topics (since 2001)
• Bathroom and toilet
• Size of workplace
• Lowest floor level
Changed questions (selected)
• Marital status – civil partnerships
• Ethnicity
– Additional tick boxes
– Revised grouping / wording
• Qualification – clearer categories
• Banded hours worked – actual hours in 2001
• Address one year ago – identification of
students
• Updated ethnicity
UK comparability
• 30 common topics
– 11 others in one or more countries
• Common wording in approx 40 out of 50
questions
Production of population estimates (1)
• Step 1: Create best possible initial address register
• Step 2: Census field work / initial questionnaire
processing
–
–
–
–
Add new addresses
Remove non-existent addresses
Assess status of non-responding addresses
Remove of duplicate responses
• Step 3: Assess and adjust for coverage
– Of addresses / households
– Of people within households, and adjust
– Using address register, census and CCS data
Production of population estimates (2)
• Step 4: Quality Assure
– a) Using “Census” data
• Visitors, second residences, dummy forms, migration data
– b) Using other sources
• Admin data, surveys, demographic data
• Step 5: Investigate
– Further visitor / second residence matching
– Further admin /demographic data analysis
• Step 6: Translate to MYE base
– 6 mths to 12 mths residence rule
– Family/permanent residence to ‘majority of time’ address
– Roll forward 3 months
• Step 7: Explain
– Second residence data (questionnaires and dummy forms)
– ?? Short term migrant data / intention to stay
Step 8: Investigate / Prepare for the
future?
• ‘Freeze’ admin datasets on 27 March 2011
• Compare census and other sources
• Especially immigration data?
Second residences in coverage
assessment / QA
• Duplicate returns from different locations:
– e.g student counted at both term-time and parents’
address, people with second residences for the
working week, children of divorced parents
• General matching process to search for
duplicates
– use second residences information to help
• During QA, use to explore address status
– No usual residents, clearly a second residence?
Second residences in 2007 postal test
• 2.5% of respondents said they had a second
residence
• A further 1% of respondents said they had a second
residence outside the UK
• Of those who said they had a second residence:
– 87% entered the address
• 69% of those entered the full postcode
• 11% of postcodes were half completed
• 20% of postcodes were left blank
• Of those that entered an address, the highest
frequency of location was London (13%), followed by
West Sussex (9%)
Visitor information for Quality
Assurance
• Sample matched back to usual residence
– check whether they were missed where they usually live
• In CCS areas, match to CCS record for household
they were visiting
– check whether they were mis-classified
– Apply the mis-classification rate to visitor numbers in
non-CCS areas
• Full match in some LAs if QA suggests concerns
• Considered matching all
– low cost benefit
– Delays
– Let the CCS do its job
Visitors in the 2001 Census
• From a sample of 7 Enumeration Areas:
– 9.5% of people recorded as visitors were actually
recorded as usual residents too
– 1.5% were of no fixed abode and should have
been recorded as usual residents
– 17% of visitors were overseas visitors
– 67% of visitors were from elsewhere in the UK
• 20% of these UK visitors were missed at their usual
residence
– 4.5% of visitors did not have an address recorded
• It is estimated there will be around 2.1 million
visitors on Census night in 2011
Use of short term migrant data
• Users will cross check census data with other sources
– 2001: used council tax lists
• 2011: likely to use information on migrants
– Already challenging MYEs on this basis
– Significant volumes (at present)
• Proposal: Migrants in UK for > 1 (3?) months fill in full
questionnaire, with ‘intention to stay question’
– Filter out 3-6 month migrants from UR base
– But use info on numbers / characteristics
• Likely to be poor quality, but plan would be to use to understand
administrative sources
– Aggregate, perhaps individual matching
“Not a census question”
• Snapshot – rapidly changing
• Quality
– Will they answer the census?
– Even if they do, would they answer an intention to
stay question
– If so, what would the quality be like?
• Would 1 month or 3 month cut off be better?
– How could we use the resulting info anyway?
Research
•
•
•
•
•
Focus groups / interviews (NCSR)
Cognitive testing
IPS interviewers
Omnibus survey
Postal test
– 10,000 random national sample of households
• 50% 6 month, no intention to stay, < 6mths as visitors
• 50% 1 month (full question set), with intention to stay
– 10,000 random sample from Northampton
• Similar 50/50 split
High level findings
• Would fill in the census, IF they realised they
had to
• Would then answer an intention to stay
question
– With a reasonable degree of confidence
– Better with a 3 month cut-off than 1 month
• Postal test
– No impact nationally
– Small, statistically significant impact locally
• Still analysing the data
Discussion
• Short term migrants
– Better to ‘ignore the problem’
• Discussion will happen post 2011
• STMs will have to make a choice
Either explicitly or implicitly
– Better to have some information than none?
• Wouldn’t provide robust estimates
• New paradigm in census questions
– include a ‘poor’ topic, solely to aid administrative source
analysis
• A step into a brave new world or a foolish
misadventure?