Family Resources Survey

1
Family Resources Survey

• Data Linking Project
• Elizabeth Whiting

2
Introduction

• Background
• Data linking and data matching
• Uses of linked data
• Pilot study
• How data will be linked
• Initial results from matching pilot data
• Next steps
• Questions?

3
The Family Resources Survey

• Launched in 1992
• 29,000 private households in UK (about 27,000 in
  GB)
• Information on incomes and benefit receipt,
  tenure and housing costs, savings
• Fieldwork carried out by ONS and NatCen

4
Background

• 2004 Strategic Review of FRS
• Problems with take-up statistics
• Improvements to administrative data
• New FRS contract from April 2006

5
Data linking and data matching

• Data Linking Linking consenting FRS respondents
  to admin data held within DWP
• Data Matching Carried out by ONS (for DWP). Data
  is matched on a one off basis and DWP given an
  anonymised dataset

6
Intended uses of linked data

• For statistical and research purposes only
• Initially to only be made available internally at
  DWP and to selected HMRC analysts
• Will not be used for operational purposes, such
  as fraud detection
• Longitudinal analyses tracking how different
  groups move in and out of work and how their
  situation changes over time
• Validation of FRS data

7
Administrative data held by DWP

• The Work and Pensions Longitudinal Study (WPLS)
  contains over 500 million lines of data covering
• Claims for Incapacity Benefit, Jobseekers
  Allowance, Income Support, Widows Benefit and
  Carers Allowance since 1999
• Claims for Bereavement Benefit, State Pension,
  DLA and AA since 2002
• Claims for Pension Credit since 2003
• Spells on New Deals, Work-based Learning etc
  since 1998
• Spells in employment from P45/46 information
  since 1998
• Annual earnings since 2003/4
• Occupational pensions in payment since 1998
• ISAs, PEPs, TESSAs, Private Pensions and Venture
  Capital Trusts as at 2001/2
• Savings accounts earning more than 50 interest
  per year from 2001/2 and updated annually

8
Informed consent

• Personal details need to be passed to DWP for
  linking (name, address, NINO, sex, date of birth)
• Requires consent of respondent (Data Protection
  Act 1998)
• Must be informed, specific and freely given
• Developed and implemented a consent question

9
2006 pilots

• Two stages
• Qualitative Cognitive testing of the consent
  question
• Quantitative Testing of the consent rate

10
Cognitive testing of consent question

• February 2006
• To test respondents understanding of the
  question
• 15 in-depth interviews
• Concluded that question needed to be shorter and
  terminology simplified

11
Quantitative pilot

• May/June 2006
• Redrafted question tested on 1,000 households
  (selected from main FRS sample)
• Consent rate of 47.5 (excluding proxy cases)
• Consent bias for some groups
• Consent falls slightly as age increases
• Consent rate lower among ethnic minorities (36
  compared to 49 for White group)
• Employees have higher consent than self-employed
• Consent question did not significantly affect the
  overall response rate to the survey
• Main reason for non-consent was concern over
  security (request for NINO thought to contribute)

12
Changes following pilot

• Copy of written question provided to respondent
  was simplified
• Question now read out by interviewer to check
  understanding
• Proxy pack revamped to make it shorter and more
  visually appealing
• Continue to request NINO
• Interviewers receive face-to-face briefing
• Consent question introduced in November 2006

13
(No Transcript)
14
Matching methodology

• Matching based on techniques used for matching
  DWP and HMRC data in WPLS
• Records are initially matched on NINO
• Hierarchical matching of other variables
• Where NINO does not match fuzzy matching is
  used
• Quality of the match is classified as by a
  traffic light system
• Continued...

15
Matching methodology

• Staged approach by NINO, then surname (soundex),
  initial of forename, DoB, gender and postcode
  sector

16
Matching methodology

• Fuzzy matching by surname (soundex), initial of
  forename, DoB, gender and postcode sector

17
Results of matching pilot data
18
Matching Rates

• Green, Amber and GreenAmber are reliable matches
  (83 of matches)
• Red and RedAmber could be improved by introducing
  further levels of matching

19
Comparison of FRS with WPLS

• Only compared benefit spells which were live at
  the time of interview.
• Receipt of following benefits were compared
  Bereavement Benefit, Carers Allowance,
  Disability Living Allowance, Attendance
  Allowance, Incapacity Benefit, Income Support,
  Job Seekers Allowance, Pension Credit, Retirement
  Pension and Severe Disability Allowance.

20
Numbers claiming benefits
21
Comparison of WPLS with FRS
22
Next Steps

• Match November 2006 data onwards to WPLS
• Carry out analysis on November 2006 data onwards
• Compare benefit amounts recorded on FRS and WPLS
• Refine the matching process to give more weight
  to certain variables
• Repeat the process using fuzzy matching only to
  assess the need to collect NINO

23
Questions

• ????

24
Consent Question

• The Department for Work and Pensions (DWP) would
  like to add your answers to information they hold
  about your benefits and tax history, employment
  and National Insurance, savings and private
  pensions.
• By adding your answers to the information that
  DWP currently hold and which they may hold in the
  future, researchers from DWP and HM Revenue and
  Customs will be able to build an accurate picture
  of how living standards in Britain change over
  time. This will allow the DWP to ensure that
  help is given where it is most needed.
• So that DWP can add your answers to their records
  we need your written permission to provide them
  with your
• Full name, Address, Sex, Date of birth, National
  Insurance Number.
• The information will be kept confidential
• Your personal details will not be passed to
  anyone else outside the team within the DWP who
  are responsible for adding your details to the
  information held by them.
• Once your answers have been added to the
  information held by the DWP, all personal details
  will be removed.
• The information will only be used for research
  and statistics by teams in DWP and HM Revenue and
  Customs.
• No one can be identified from the research, as
  personal details are never included in the
  results.
• Any current or future claim for benefits or tax
  credits will not be affected. The information
  will not be used to work out whether anyone is
  claiming benefits or tax credits they should not
  be.
• You can withdraw your permission to add your
  information to DWP records at any time and your
  linked information will be removed from all
  future DWP records.