Constructing population time series with an ethnic breakdown for sub-national areas in England and Wales, Albert Sabater PhD student at CCSR University of Manchester CCSR Seminars th March 2007
Motivation Constructing population time series with an ethnic breakdown for sub-national areas in England and Wales, Can we say that Yorks Indian population is growing without taking into account boundary changes? -Can we say that Birminghams Caribbean population is ageing without taking into account different treatment of undercount? -Can we compare published statistics by ethnic group from 1991 and 2001 census output without adjustments?
Summary 1.Aim 2.Problems 3.Research strategy 4.Some results 5.Dissemination of data Constructing population time series with an ethnic breakdown for sub-national areas in England and Wales,
1.Full population estimates for 1991 and 2001, age, sex, ethnic group, using 2001 Census boundaries 2.Analysis of population change between 1991 and 2001 Detail:Age: 91 categories Sex: 2 categories Ethnic group: 10 (1991) and 16 (2001) Aim Mid-1991 small area population estimates Districts Wards Output Areas Mid-2001 small area population estimates Districts Wards Output Areas
Problems Problem 4 Boundary changes Problem 3 Ethnic group and age categories Problem 2 Treatment of non-response Problem 1 Population definition -Enumeration of students -Timing adjustment
Data linked: Censuses Geographies: wards and levels above in E&W and Scotland Correction for students and timing transfer Allowance of non-response (missing million from EwC) Ethnic groups: White, Asian, Black and Other Previous work: Linking Censuses Through Time (LCT) LCT constraints by ethnic group: Not cross-tabulated by age and sex Non-response not updated with latest ONS MYEs Non-response not differential by ethnic group
Revising 1991 MYEs without ethnic group 1991 EwC MYE = census + undercount + timing adjustment + student transfer From 1991 Enumeration Districts to 2001 Output Areas STEP 1 PART 1.EwC data (EWCPOP for EDs)
Boundary change Source geographyTarget geography No (direct allocation) Yes (multiple allocation) Yes (multiple allocation with intersection) Area 1 Zone 1 Zone 2 Area 2 Area 3 Area 4 Area 5 Zone 3 Zone 4 Weight *0.6 *0.4 Using Geographical Conversion Tables
Revising 1991 MYEs without ethnic group PART 1.EwC data (EWCPOP for EDs) & ONS data (revised MYEs for Districts) 1991 EwC MYE = census + undercount + timing adjustment + student transfer From 1991 Enumeration Districts to 2001 Output Areas STEP 1 Aggregation to 2001 DistrictsONS revised 1991 District MYE Revision of 1991 undercount STEP 2
Revision of 1991 undercount after 2001 Downwards -350K
Revising 1991 MYEs without ethnic group 1991 EwC MYE = census + undercount + timing adjustment + student transfer From 1991 Enumeration Districts to 2001 Output Areas STEP 1 Aggregation to 2001 DistrictsONS revised 1991 District MYE Revision of 1991 undercount STEP 2 Derive single year of age STEP 3 PART 1.EwC data (EWCPOP for EDs) & ONS data (revised MYEs for Districts)
Revising 1991 MYEs by ethnic group 1991 Census = SOCPOP PART 2.EwC data (SOCPOP for wards) & Previous estimate (PART 1 for OAs) Share SOCPOP from 1991 wards to 1991 Enumeration Districts STEP 1 From 1991 Enumeration Districts to 2001 Output Areas STEP 2Scale to 1991 revised MYE without ethnic groupSTEP 3Add student adjusment STEP 4
OPTION 1 Addition & Removal: 2001 term-time students OPTION 2 Addition & Removal: 1991 resident population OPTION 3 Addition & Removal: Net impact only on White students OPTION 4 Addition: 2001 term-time students Removal: 1991 resident population 1991 ethnic group student adjustment Total Net impact = Students added + Students removed Using EwC data (by age and sex for EDs) Ethnic group dimension
Percentage distribution between ethnic groups of those aged Source: 1991 SAR
Revising 1991 MYEs by ethnic group 1991 Census = SOCPOP PART 2.EwC data (SOCPOP for wards) & Previous estimate (PART 1 for OAs) Share the SOCPOP from 1991 wards to 1991 Enumeration Districts STEP 1 From 1991 Enumeration Districts to 2001 Output Areas STEP 2Scale to 1991 revised MYE without ethnic groupSTEP 3Add student adjusment STEP 4 Derive single year of age STEP 5 From 2001 Output Areas to wards and Districts STEP 6
2001 MYEs by ethnic group 2001 One Number Census Timing adjustment ONS data from experimental statistics (for Districts) STEP 1
2001 timing – England 9 weeks April 29 to June 30 Births DeathsMigration
2001 MYEs by ethnic group 2001 One Number Census Timing adjustment ONS data from experimental statistics (for districts) & non-response imputation rates STEP 1 Extra non-response ONS imputation rates STEP 2
2001 extra non-response using ONS imputation rates
2001 extra non-response in Birmingham ONS imputation ratesONS experimental statistics
2001 ST wards Output Areas 2001 One Number Census Timing adjustment Extra non-response ONS imputation rates STEP 4 STEP 1 STEP 2 STEP 3 ONS data from experimental statistics (for districts) & non-response imputation rates 2001 MYEs by ethnic group
Population change for Districts in E&W Censuses 1991 and 2001 as publishedFull estimate with 2001 boundaries Note: Base map created by Danny Dorling and Helen Durham
Population change Birmingham
Detail:Age: 91 categories Sex: 2 categories Ethnic group: 10 (1991) and 16 (2001) Mid-1991 small area population estimates Districts Wards Output Areas Mid-2001 small area population estimates Districts Wards Output Areas Mid-2001 small area population estimates Districts Wards Allocation of extra non-response from ONS experimental statistics Allocation of extra non-response using ONS imputation rates Data summary and dissemination Work in progress