More Related Content
What's hot (20)
Viewers also liked (15)
Similar to Evaluating the feasibility of using administrative data in the context of census population statistics (20)
Recently uploaded (20)
Evaluating the feasibility of using administrative data in the context of census population statistics
- 1. Evaluating the feasibility of using administrative data in the context of census population statistics Kimberley Brett Office for National Statistics Census Transformation Programme Beyond 2021 Research and Design
- 2. Overview • Use of administrative data for population statistics requires careful consideration of quality issues - Coverage - Lags - Definitions - Policy - Administrative and operational process • Outline how we are developing a framework for evaluating and improving quality of administrative data used in SPD construction • Examples of quality issues in our analysis
- 3. Admin Source Reports Beyond 2011 • During phase 1 Beyond 2011 published a series of reports on the following data sources • Patient Register (PR) • Customer Information System (CIS) • Higher Education Statistics Agency (HESA)-Student Records • English School Census and Welsh School Census • Electoral Register • Evaluation framework that focused on each source separately: • Coverage • Plausibility • Metadata • Understanding administrative process
- 4. Assessing Coverage – Comparison with Census Estimates • Patient Register 2011 • Customer Information System 2011
- 5. Assessing Coverage – Comparison with Census Estimates • School Census • Electoral Register
- 6. Developing the quality framework • Our approach is to combine administrative sources to estimate the population and its characteristics • Need to evolve the quality framework to identify and improve on specific issues emerging in the SPD analysis • Stability of the admin data • Source conflicts • Accounting for missing groups • Established a Data Suppliers Group with government departments supplying the data: • Acquisition of additional data to supplement SPD • Feedback on the quality issues identified in our analysis • Try and influence how data is collected and maintained for statistical purposes • UKSA quality toolkit provides us with a framework to work towards
- 7. SPD 5 population counts compared to the 2011 Census 91% of LA total population counts within 3.8% of Census estimate in 2011 Admin data method lower than 2011 Census Admin data method higher than 2011 Census 7
- 8. SPD 5 population counts compared to the 2011 Census • 20- 24 year old males
- 9. Using linked data to evaluate quality • Linking data across multiple administrative sources and Census data provides valuable insight to explain differences between SPD counts and Official Estimates • Can measure the accuracy • How long lags persists for cohorts of the population • Definitional and collection mode differences • Volatility of operational processes
- 10. Lags of a Statistical Population Dataset • Counts on the SPD correlate well with Census estimates, but lags exist in particular locations Age by single year Percentage of the population by single year of age that are recorded in the same location when comparing the SPD to census estimates
- 11. Lags on the Patient Register (PR) • How long does it take for PR records to update location? Same address information on 2011 PR as Census Same address information on 2012 PR as Census Same address information on 2013 PR as Census Differentaddress information on 2011 PR, 2012 PR and 2013 PR to Census
- 12. Examples of changes in operational processes • The National Duplicate Registration Initiative (NDRI) was the Audit Commission’s periodic exercise that used data matching techniques to review GPs’ patient lists • Conducted list cleaning in 1999, 2004 and 2009 Removal of FP69s • 2004 - 185,000 patient registration deductions • 2009 - 95,000 patient registration deductions • At present, this is done on an ad hoc basis by NHS areas • Subsequently, in some LAs we observe significant reductions of patient numbers following list cleaning exercises that result in decreases of population counts on SPDs
- 13. Example of list cleaning and impact on SPD count 13
- 14. Example of operational processing and impact on SPD count 0 1000000 2000000 3000000 4000000 5000000 6000000 7000000 1 2 3 4 5 6 7 8 9 10 11 12 Population Month
- 15. Example of operational processing and impact on SPD count 0 500000 1000000 1500000 2000000 2500000 3000000 3500000 0 5 10 15 20 25 30 35 Population Day of the month
- 16. Quality of admin data for household statistics • Census collects information at household level: • Definition of “household” relates to ... • Household definition not captured on admin data • Limited to an address definition when producing household statistics • Aims to produce household statistics in 2016: • Number of households • Household size • Household composition • Challenges in producing these statistics: • Unable to geo-reference all addresses to an address frame • Churn at the address results in inflated household sizes • Relationships between family members not recorded on admin data
- 17. Missing Address Identifiers (OSAPRs)
- 19. Modes of collection • Later releases of Research Outputs will explore potential of population characteristics. • Will depend on data availability and quality. • Potential topics include: • Income • Ethnicity • Health • Need to understand how the definitional differences and modes of collection impact on the statistics produced from admin data. • Undertaken analysis of linked records between the 2011 Census and School Census to compare ethnicity.
- 20. 2011 Census ethnicity English School Census ethnicity WhiteBritish Irish IrishTraveller/ Gypsy/Romany Indian Bangladeshi Pakistani Whiteand Asian OtherAsian Chinese African Whiteand BlackAfrican Caribbean Whiteand Black Caribbean OtherWhite OtherBlack OtherMixed OtherEthnicity Missing Total (denominator) White British 95% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 2% 0.50% 0.50% 0.50% 2% 5,048,672 Irish 41% 47% 1% 0.50% 0.50% 0.50% 1% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 5% 0.50% 3% 0.50% 2% 22,609 Irish Traveller/ Gypsy/Romany 35% 2% 54% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 6% 0.50% 0.50% 1% 2% 9,150 Indian 0.50 % 0.50% 0.50% 89% 0.50% 1% 1% 5% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 1% 0.50% 2% 169,609 Bangladeshi 0.50 % 0.50% 0.50% 0.50% 92% 1% 0.50% 2% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 4% 99,905 Pakistani 0.50 % 0.50% 0.50% 1% 0.50% 86% 1% 4% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 1% 3% 4% 252,189 White and Asian 11% 0.50% 0.50% 1% 0.50% 2% 54% 3% 0.50% 0.50% 0.50% 0.50% 0.50% 3% 0.50% 15% 3% 4% 82,152 Other Asian 1% 0.50% 0.50% 12% 0.50% 2% 2% 58% 1% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 4% 17% 2% 84,028 Chinese 2% 0.50% 0.50% 0.50% 0.50% 0.50% 1% 2% 83% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 7% 2% 2% 27,577 African 1% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 1% 0.50% 83% 1% 1% 0.50% 1% 7% 2% 1% 3% 190,489 White and Black African 6% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 8% 55% 1% 3% 3% 3% 14% 2% 4% 38,611 Caribbean 1% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 3% 0.50% 77% 3% 0.50% 9% 3% 1% 4% 71,256 White and Black Caribbean 12% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 2% 3% 62% 1% 2% 12% 1% 4% 108,920 Other White 8% 0.50% 1% 0.50% 0.50% 0.50% 1% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 75% 0.50% 6% 5% 3% 169,626 Other Black 1% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 0.50% 34% 1% 30% 2% 0.50% 20% 6% 1% 4% 27,625 Other Mixed 10% 0.50% 0.50% 1% 0.50% 1% 5% 3% 1% 2% 3% 2% 8% 5% 4% 47% 5% 5% 23,763 Other Ethnicity 5% 0.50% 0.50% 0.50% 0.50% 0.50% 2% 12% 0.50% 2% 1% 0.50% 0.50% 10% 2% 10% 50% 4% 66,760 Missing 69% 0.50% 0.50% 2% 1% 3% 1% 2% 0.50% 4% 0.50% 2% 1% 5% 1% 2% 2% 3% 222,193 Census and England School Census Ethnicity
- 21. Admin data quality and how it impacts SPDs 21 Risk / Issue Example Mitigation Definitional differences Definition of residence Definition of household Lack of common address identifier e.g. (UPRN) Harmonisation, combined use with surveys Single address register and UPRN at source Registration People not de- registered when abroad or dead Not registering on arrival or moving lags in registration and update Using activity data and combined survey use Feedback to data suppliers Policy or operational change Patient register list cleaning Benefit change, e.g. Universal credit Consultation and engagement with ONS/GSS, Through new legislation?
- 22. Summary – The potential future of administrative data • Need to incorporate more evidence of activity from admin data to help improve quality of SPDs • More research to understand whether there are different characteristics of those people who are included or are not included on the SPD • Need to maintain and establish positive relationships with Data Suppliers
- 23. References • Source reports for administrative datasets can be found on the ONS website under: http://www.ons.gov.uk/ons/about-ons/who-ons-are/programmes- and-projects/beyond-2011/reports-and-publications/index.html
- 24. Census Transformation Programme Annual Research Conference Covering: 2021 Census design, Census topic consultation, linking admin data, research outputs 26-27th November 2015 Chichester College To register an interest or for more information please email: 2021census.consultation@ons.gov.uk