towards an integrated household survey nikki bennett office for national statistics

26
Towards an integrated household survey Nikki Bennett Office For National Statistics

Upload: gabriel-powell

Post on 28-Mar-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Towards an integrated household survey Nikki Bennett Office For National Statistics

Towards an integrated household survey

Nikki Bennett Office For National Statistics

Page 2: Towards an integrated household survey Nikki Bennett Office For National Statistics

Overview

• Aims and benefits• Key design features• Consultation • Survey content

– core module– Other modules (sub-sample topic-specific modules)– sample sizes for analysis purposes

• Outputs – aggregate outputs and survey microdata • Future consultation• Development timetable

Page 3: Towards an integrated household survey Nikki Bennett Office For National Statistics

Aim

To develop an integrated household survey that will:

– make better use of data already collected

– deliver a range of better quality, more reliable estimates at national, regional and particularly, sub-regional levels

– produce a range of new, regular outputs from a very large dataset of core information

– improve coherence in official statistics through fewer competing survey estimates

– deliver efficiencies through standardisation, integration, modernisation and better design

Page 4: Towards an integrated household survey Nikki Bennett Office For National Statistics

ONS surveys for integration

• Labour Force Survey (LFS)– including annual sample boosts

• General Household Survey (GHS)• Expenditure & Food Survey (EFS)• Omnibus Survey (OMN)

• Annual Population Survey boost sample no longer included

• Scope for other surveys to join after launch year

Page 5: Towards an integrated household survey Nikki Bennett Office For National Statistics

Statistical benefits (1)

• greater coherence in official statistics, fewer ‘competing estimates’ between surveys because of data from one combined source

• better quality, more reliable estimates at national, regional and particularly, sub-regional levels in between each decennial Census

• a range of new, regular outputs from a very large core sample – e.g. banded household income

Page 6: Towards an integrated household survey Nikki Bennett Office For National Statistics

Statistical benefits (2)

• Greater harmonisation of concepts, classifications, questions and outputs

• Increased flexibility to sample respondents to produce estimates to required level of precision

• Better sample coverage at local authority district level through reorganisation of sample and fieldwork

• More responsive and flexible in meeting new needs – can add a new topic module or new interview combination – and much cheaper than a new survey

Page 7: Towards an integrated household survey Nikki Bennett Office For National Statistics

Key design features

• unclustered sample of addresses• core set of information collected from all selected

households followed by a particular interview type covering specific topics

– once the sample is selected each address will be allocated to a particular interview type

• single integrated field force of interviewers able to administer all types of interview combinations in small geographical areas

• an enhanced survey case management system to increase our ability to manage field operations effectively and efficiently

Page 8: Towards an integrated household survey Nikki Bennett Office For National Statistics

Modular approach

• a single modular questionnaire instrument

• a core set of information covering all households and adult members;

• ‘Fixed’ content – 5 years +• ‘Rotating’ content

– topic modules of questions administered to sub-sets of the sample

– core and topic modules combined to create a number of different interview types

Page 9: Towards an integrated household survey Nikki Bennett Office For National Statistics

Fixed core Rotatingcore

Topic A Topic B

Topic C Topic D

Topic C Topic E

Sam

ple

siz

e

Interviewcombination

X (panel)

Interview length

Interviewcombination

Y(X-sectional)

Interviewcombination

Z(diary

component)

Abstract representation of the modular design

Page 10: Towards an integrated household survey Nikki Bennett Office For National Statistics

Existing sample sizes

Achieved number of independent household interviews per annum - (Great Britain)

LFS 85,000

LFS annual boosts 88,000

EFS 6,000

GHS 9,000

Omnibus 16,000

All 204,000

Plus Northern Ireland data for UK coverage

Page 11: Towards an integrated household survey Nikki Bennett Office For National Statistics

Panel elements of the integrated design

Survey Number of interviews

Interval between interviews

Time in sample

QLFS 5(address panel)

3 months 12 months

Local area LFS boosts

4(address panel)

12 months

4 years

GHS 4 (household

panel)

12 months

4 years

Page 12: Towards an integrated household survey Nikki Bennett Office For National Statistics

Consultation process

• During 2004: – bi-laterals with existing funding departments– talks to range of audiences – central and local government,

academia, SRA, various international conferences

• October 2004: ONS published a consultation document on the NS website (100+ responses)

– 25% central govt– 40% local or regional govt and Primary Care Trusts– 14% special interest groups– 8% academia– 12 % other (business or individuals)

• October 2005: Consultation response document

Page 13: Towards an integrated household survey Nikki Bennett Office For National Statistics

Criteria for inclusion in core module

• Classificatory variable essential for analysis• Clear requirement for a high level of precision

nationally, regionally or sub-regionally and not provided elsewhere

• Information that can be collected either face-to-face or by telephone interview

• Proxy responses acceptable• Topics which would not adversely affect response

to the interview as a whole• Stability of funding over time

Page 14: Towards an integrated household survey Nikki Bennett Office For National Statistics

Potential core outputs

• People: sex, age, marital status (incl. civil partnership status)

• Identity: ethnicity, religion, national identity, country of birth, year of entry to UK

• Living arrangements: e.g with partner

• Household type and composition (incl. family type)

• Family Units (within households) e.g. couple with dependent children

Page 15: Towards an integrated household survey Nikki Bennett Office For National Statistics

Potential core outputs

• Usual place of residence

• Residence one year ago

• Health: general health (healthy life expectancy derived), limiting long-term illness (disability free life expectancy), prevalence of smoking

• Housing: tenure, household size, accommodation type

• NS-SEC

• Income: banded household income (equivalised quintile groups)

Page 16: Towards an integrated household survey Nikki Bennett Office For National Statistics

Potential core outputs

• Employment-related: economic activity status (derived), self-employed or employee, supervisor or not, full-time/part-time, occupation & industry group, place of work, no. of employees at workplace, whether looking for paid work, govt schemes

• Education: educational status (whether student), highest qualification level, age finished full-time education

• Access to car (or van)

Page 17: Towards an integrated household survey Nikki Bennett Office For National Statistics

Summary of potential core outputs

Sex

Age

Family type and composition

Health: general health, limiting long term illness, smoking

Employment:

Economic & employment status, occupation and industry

Marital status

Living arrangements

Identity:

Ethnicity,

Religion, national identity,

Country of birth, yr to UK

Housing: tenure, hhld size, accommodation type

Education: educational status, highest qualification, age left f/t education

Household type and composition

Residence: usual, one year ago

NS-SEC

Household income

Access to car or van

Page 18: Towards an integrated household survey Nikki Bennett Office For National Statistics

Core outputs: examples of approximate households for analysis (per annum)

204 000 households (per annum)

• 63 000 one person households• 14 000 households of a lone parent with dependent

children• 6 000 households of a cohabiting couple with

dependent children• 41 000 households in social sector housing• 22 000 privately renting households

Page 19: Towards an integrated household survey Nikki Bennett Office For National Statistics

Core outputs: examples of approximate numbers for analysis (per annum)

450 000 persons (all ages)

• 328 000 people aged 16 and over• 45 000 people aged 16-24• 18 000 people aged 75 and over

Ethnic group• 7000 Indian • 7000 Pakistani • 2000 Bangladeshi • 10 000 Black

Page 20: Towards an integrated household survey Nikki Bennett Office For National Statistics

Topic modules

• existing survey content rearranged into different interview types

• currently trialling 5 different interview types but this may change

• dependent on departmental needs and priorities and availability of appropriate funding

• scope for– Adding topic modules to existing interview types

(subject to interview length)– Adding new interview type if adequate

demand/funding

Page 21: Towards an integrated household survey Nikki Bennett Office For National Statistics

Combining years of data

Combining years of IHS data: likely number of different households (excl. LFS boosts)

• Core module:- two years = 185 000; three yrs = 252 000

• Labour market data: – two yrs = 128 000; three yrs = 170 000

• GHS-specific data– two yrs = 11250; three yrs = 13500

• EFS-specific data- Two yrs = 12 000; three yrs = 18 000

Page 22: Towards an integrated household survey Nikki Bennett Office For National Statistics

Future consultation

• Qtr 1 2006: consultation with existing/potential funders - data requirements, continuity issues, funding

• Qtr 2-3 2006: – report on Pilot 1 (fieldwork: Feb-March 2006).– discuss options and recommendations for future IHS content

and modular design (topics and interview combinations) – Discuss future funding arrangements and charging mechanism

(dependent on cost modelling based on information from Pilot 1)

• Qtr 4 2006: 'shortlist of topics and funders' for IHS 2008 and 3-5 year forward plan

• By end of Qtr 1 2007: agree core and topic content for IHS 2008

Page 23: Towards an integrated household survey Nikki Bennett Office For National Statistics

Outputs

• Continue to deliver range of topic-specific outputs currently produced on existing surveys and new topic outputs (subject to availability of funding and resources): monthly, quarterly, annual

• Range of regular outputs from core module– Annual datasets via UKDA;– aggregate tables via Web (NOMIS/NeSS)– Quarterly production of aggregate tables on a

rolling annual basis (Jan-Dec; Apr-March; July-June; Oct-Sept; Jan-Dec…)

Page 24: Towards an integrated household survey Nikki Bennett Office For National Statistics

Microdata

• Proposal for three types of microdata:– Detailed but anonymous dataset to GSS users with

low levels of geography;– As above, but to trusted users, outside GSS, via

UKDA under Special Licence. Proposed that this will have UA/LA codes connected to dataset and will provide richer source of data than existing annual local area LFS data

– A non-disclosive public dataset for release via UKDA. No geography below GOR level. Release of public data to coincide with release of tabular data to NeSS (social tables) and NOMIS (labour market tables)

Page 25: Towards an integrated household survey Nikki Bennett Office For National Statistics

Development timetable

Completed

• 2004: initiate project - build up team, develop plans• July 2004: Publication of initial formal consultation document (3

months)• Autumn 2004-Spring 2005: develop modular design • Spring and summer 2005: small scale field trials

To be done

• 2006 - mid 2007: fieldwork modernisation and field force integration; two pilots – one with improved case management system

• January 2008– IHS to start – Parallel run LFS with IHS

Page 26: Towards an integrated household survey Nikki Bennett Office For National Statistics

Thank you