Introduction to the Integrated Household Survey (IHS) Richard Clarke Caroline Jones ONS Social...

Preview:

Citation preview

Introduction to the Integrated Household Survey (IHS)

Richard Clarke

Caroline Jones

ONS Social Surveys

Outline

• Background of the IHS

• How do these Social Surveys work then?

• Datasets – what and when!

• What are the results from the IHS?

Outline cont’d

• Dataset variables – End User License (EUL) vs. Special License (SL)

• Analysing the data – what to watch out for

• Questions and Answers

Background to the IHS

• ONS has several separate household surveys

Different samples, fieldwork designs and processing systems

• All surveys have similar introductions

Names, ages, work history

• Aim of the IHS is to integrate surveys into one modular survey with a core set of questions and survey modules ‘bolted’ to the core

Leading to …..

• The largest social survey ever produced by ONS

Why ONS wanted to Integrate its Surveys

• Harmonisation of questions

IHS consolidates existing work to harmonise questions over existing surveys

• Inter-Censal Data

pressure to provide more accurate data between censuses

• Demand for Local Level Data

increasing demand for more information, particularly at a local level within the UK

Surveys making up the IHS in 2009/10

• Annual Population Survey (APS) 4 wave panel interview at 12 month intervals + waves 1 + 5 of

the Quarterly LFS

• Living Costs and Food Survey (LCF)

single interview. Two week diary-keeping element.

• General Lifestyle Survey (GLF) 4 wave panel interview at 12 month intervals - longitudinal

• Life Opportunities Survey (LOS) a longitudinal survey

• English Housing Survey (EHS) single interview + surveyor visit

• Opinions Survey (OPN) single interview (OPN left the IHS after Dec. 2010)

Surveys making up the IHS in 2009/10 cont’d

IHS: 2009/ 2010 Sample Size by Component Survey

Interview Type April 2009 to March 2010 Achieved

Interviews

GLF 18,000

LCF 12,000

OPN 21,000

EHS 41,000

APS +LFS 334,000

LOS 23,000

Annual Total 449,000

•The expected breakdown of the annual sample is as follows:

• Common set of core questions

• Continual sampling

• Adding up to large scale quarterly datasets.

What is The Survey Process?

Design

PrepareTest

Conduct Survey / Collect Data

Assemble andCheck Data

Disseminate Data

Development of the questions

• Most IHS questions were pre-existing ones originating on the APS/LFS

• Sponsored questions added: e.g. for smoking prevalence

• New questions undergo extensive checks – e.g. cognitive testing and field tests before going live

• Consult with stakeholders

• Program the questionnaire

What are the IHS Core Questions

• Household composition• Accommodation• Tenure• Nationality• Country of Birth• National identity• Ethnicity• Sexual identity• Religion

• Period at current address

• Health• Smoking• Government training

scheme• Work• Looking for work• Education

Find someone to ask it to….

• The Royal Mail small user Postcode Address File (PAF) is the main sampling frame

We only interview people living in private households

Students in halls of residence or boarding

schools should be included at their parents address

Communal establishments are largely excluded

• Many are chosen.. Sadly not everyone will respond

Collecting the Data

• Data collected by CAPI

(Computer Assisted Personal Interviewing)

• 65% of interviews by field interviewers

in respondents home, remainder (mainly APS) done by telephone unit

• Respondents can answer for others in the household = Proxy Interviews

• Around 40% of IHS data collected in this wayException is sexual identity which is only asked in person

An example of what the interviewer sees

An example of a showcard used by interviewers

Please choose all that apply

1. Scottish

2. English

3. Welsh

4. Northern Irish

5. British

6. Other (please describe)

2 S

Data checks are built into the Questionnaire

Later, back in Newport

• The IHS team gets monthly data files from each survey

• Files are combined to give annual dataset

• Data cleaningMake sure all variable names are the same!

Harmonise the questions

e.g. GLF and APS have extensive education sections which

are simplified into one highest qualification variable for the

IHS

What Datasets are available to you

• IHS releases data quarterly with a full years dataApril 2009 to March 2010 July 2009 to June 2010 October 2009 to September 2010 etc.

• Two type of publicly available datasets

End User License (EUL)

Special License Dataset (SL)

End User License vs. Special License

Special License• More disclosive

personal data, e.g. sexual identity, relationship grid

• Smaller Geographic areasDown to local authority

level

• Users need to apply & be accepted for approved researcher status

End User License• No disclosive

information

• Large geographic areasDown to Government

Office Regions

• Available after registration on ESDS website

Local authorities, counties

Examples of Variables ….

SLCountryGOR

Geography

Sex, ageEthnicity, marital status, national identity, religion,5 nationality categories, country of birth (3 categories UK, EU, other)

Relation to other people in household, sexual identity, civil partnerships, other nationality categories(200+), country of birth (200+ countries)

Identity SL

EUL

EUL

Variables cont’d

Industry & occupation codes

Economic activity (top coded)

Employment/training status

Detailed levels of occupation/industry coding

Work

Many more variables (197 on SL file) – see User Guides on the IHS Website

www.ons.gov.uk

EUL

SL

Analysing the IHS Data

Points to bear in mind:• Data has been imputed

Missing data “filled in” using nearest neighbour approach

• WeightingWeights to equate numbers of responses to the population

total• Household Weight – for most analysis

• Non-Proxy weight – for analysing sexual identity(Will be Sexual Identity Weight on subsequent

datasets)

• Sampling ErrorsWith any sampling survey there are confidence limits on the

data• Given in appendix 2 of the IHS bulletin

Example Output 1Results for GB and constituent countries – General Health, males aged 16+, Apr 2009- Mar 2010

42.7% 41.2% 43.3% 42.7%

37.7%34.6% 35.6% 37.4%

14.0%16.8% 14.4% 14.2%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

England Wales Scotland GB

Very badBadFairGoodVery good

Source: IHS

Example Output 2Dichotomised ‘Good’ General Health, adults aged 18-64 by Economic Position, Wales Apr 2009- Mar 2010

75.7%

50.1%

82.1%88.9% 74.0%

59.8%

78.8%88.4%

25%30%35%40%45%50%55%60%65%70%75%80%85%90%95%

In employment ILO Unemployed Inactive Good health' inWales

Male Female

Source: IHS

Media coverage

The Daily Mirror

The Sun

City AM

IHS website from www.statistics.gov.uk

Summary

• The IHS is the largest ONS Social SurveyMade up from a common core on 6 other ONS surveys

• Annual Datasets updated every 3 months

• End User License File for higher level analysis• Special User License File for more in-depth

investigations

• See the IHS Website for User Guides www.ons.gov.uk

Any Questions?

Richard.Clarke@ons.gsi.gov.uk

Caroline.Jones@ons.gsi.gov.uk

Recommended