Introduction to methodological issues in LS ethnicity research Julian Buxton Bola Akinwale Issues • Sample sizes • Item imputation in 2001 • Change in recorded ethnic group, 1991-2001 • Loss to follow-up Ethnicity in the LS • 1971 Census: Country of birth, parental COB • 1981 Census: Country of birth • 1991 Census: Ethnicity, Country of birth • 2001 Census: Ethnicity, Country of birth and religion (voluntary) Table 1 – Ethnic group at 2001 (all people) 2001 Census LS traced sample N % WHITE White British 467,822 White Irish 6,311 Other White 12,835 MIXED White & Black Caribbean 2,384 White & Black African 780 White & Asian 1,882 Other Mixed 1,496 ASIAN or ASIAN BRITISH Indian 12,487 Pakistani 8,350 Bangladeshi 3,277 Other Asian 2,606 BLACK or BLACK BRITISH Black Caribbean 4,837 Black African 3,993 Other Black 926 CHINESE or OTHER ETHNIC GROUP Chinese 2,213 Other Ethnic Group 2,174 TOTAL 534,373 2001 ONC population N % 87.5 1.2 2.4 45,533,741 641,804 1,345,321 87.5 1.2 2.6 0.4 0.1 0.4 0.3 237,420 78,911 189,015 155,688 0.5 0.2 0.4 0.3 2.3 1.6 0.6 0.5 1,036,807 714,826 280,830 241,274 2.0 1.4 0.5 0.5 0.9 0.7 0.2 563,843 479,665 96,069 1.1 0.9 0.2 0.4 0.4 100.0 226,948 219,754 52,041,916 0.4 0.4 100.0 Source: Table S01.6 2001 LS Sampling Fractions by sex and ethnic group Table 2 – Religion at 2001 2001 LS traced sample 2001ONC population Christian Buddhist Hindu Jewish Muslim Sikh All other religions No religion Religion not stated N 384,462 1,486 6,251 2,632 17,586 4,614 1,501 76,093 39,748 % 71.9 0.3 1.2 0.5 3.3 0.9 0.3 14.2 7.4 N 37,338,486 144,453 552,421 259,927 1,546,626 329,358 150,720 7,709,267 4,010,658 % 71.7 0.3 1.1 0.5 3.0 0.6 0.3 14.8 7.7 TOTAL 534,373 100.0 52,041,916 100.0 Source: Table S01.5 2001 LS Sampling fractions by sex and religion Table 3 – Country of birth at 2001 England Scotland Northern Ireland Wales Rest of Europe AFRICA ASIA NORTH AMERICA SOUTH AMERICA OCEANIA Other TOTAL 2001 LS traced sample N % 445,054 83.3 8,028 1.5 2,221 0.4 28,570 5.3 15,154 2.8 2001 ONC population N % 43,566,005 83.7 818,966 1.6 222,975 0.4 2,798,465 5.4 1,491,939 2.9 7,866 20,893 4,088 757 1,288 454 1.5 3.9 0.8 0.1 0.2 0.1 808,940 1,594,302 466,995 73,785 158,488 46,347 1.6 3.1 0.9 0.1 0.3 0.1 534,373 100.0 52,047,207 100.0 Source: Table S01.4 2001 LS Sampling fractions by sex and country of birth Table 4 – Ethnic group at 1991 (all people) 1991 LS traced sample White Black Caribbean Black African Black Other Indian Pakistani Bangladeshi Chinese Other Asian Other N 499,853 5,220 2,001 1,723 11,180 6,104 2,385 1,641 2,071 2,837 % 93.4 1.0 0.4 0.3 2.1 1.1 0.4 0.3 0.4 0.5 Total 535,015 100 Source: Table L9101.7 Forward linkage rates between 1991 and 2001 Censuses Table 5 – Ethnic group at 2001 by sex and age 0-15 White Mixed Black Caribbean Black African Black Other Indian Pakistani Bangladeshi Chinese Other MALE 16-59 47,496 138,743 1,616 1,434 474 1,287 582 1,194 179 239 1,112 4,063 1,191 2,642 427 943 203 755 517 1,659 TOTAL 53,797 152,959 Source: ONS Longitudinal Study FEMALE 16-59 60+ 0-15 60+ 48,576 136 476 121 29 928 467 187 129 218 45,762 143,312 1,584 1,610 453 1,658 502 1,490 157 302 1,097 4,456 1,122 2,653 464 1,165 181 817 463 1,748 63,083 162 490 104 20 832 275 91 129 175 51,267 258,023 51,785 159,211 65,361 276,357 Table 6 – Ethnic group by NSSEC at 2001 (16-59 year olds) NSSEC at 2001 Higher Lower Prof Intermed Employers Own Man & Prof & Higher Tech in small orgs Account White Mixed Black Caribbean Black African Black Other Indian Pakistani Bangladeshi Chinese Other Lower tech & super Semi Routine Never worked FT routine & l-t u/e students 27,790 288 207 245 33 986 280 68 188 359 65,075 561 666 454 110 1,155 426 120 231 576 34,180 323 433 234 55 794 288 88 99 263 7,941 53 41 45 5 446 224 115 159 133 14,964 101 89 58 18 422 321 21 73 113 26,148 187 227 137 36 486 229 119 54 194 42,938 396 442 399 90 1,230 657 360 186 462 34,146 277 315 196 60 1,239 713 239 77 290 9,038 273 207 354 45 713 1,531 742 102 472 19,805 584 316 562 89 1,048 625 236 403 545 TOTAL 30,444 Source:ONS Longitudinal Study 69,374 36,757 9,162 16,180 27,817 47,160 37,552 13,477 24,213 Table 7 – Ethnic group by Economic activity at 2001 (16-59 year olds) Economic Activity at 2001 Employed Employed Self- Seeking Retired Student Looking Perm sick FT PT employed work after home White Mixed Black Caribbean Black African Indian Pakistani Bangladeshi Chinese Other Other 138,874 1,142 1,400 968 3,471 1,204 292 505 1,250 39,497 280 310 233 860 404 273 130 315 26,102 199 149 134 975 494 125 236 279 9,768 204 217 236 391 388 165 44 182 4,071 11 30 17 89 30 11 20 23 20,265 613 335 610 1,081 660 248 408 585 21,156 276 201 234 743 1,219 599 135 411 14,396 135 164 84 442 322 108 27 117 7,896 183 137 168 467 573 287 67 245 TOTAL 149,106 Source: ONS Longitudinal Study 42,302 28,693 11,595 4,302 24,805 24,974 15,795 10,023 Table 8 – Ethnic groups by Government Office Regions at 2001 (all people) North East White Mixed Black Caribbean Black African Black Other Indian Pakistani Bangladeshi Chinese Other Government Office Regions at 2001 North Yorkshire East West East of London South South Wales West & Humber Midlands Midlands England East West 25,344 65,084 153 607 11 172 35 139 4 62 136 803 142 1,447 62 336 42 222 77 306 48,421 478 241 96 44 617 1,755 165 141 214 40,612 48,748 52,757 468 693 548 220 785 246 75 108 128 30 90 37 1,578 2,457 667 339 1,777 501 106 399 232 160 171 252 210 396 292 48,216 78,597 50,334 28,859 2,157 850 389 199 2,785 222 130 26 3,060 247 68 37 574 54 25 6 4,858 1,094 185 93 1,474 747 66 102 1,677 178 43 79 706 328 130 62 2,467 586 149 83 TOTAL 26,006 69,178 Source: ONS Longitudinal Study 52,172 43,798 55,624 55,660 67,974 82,903 51,519 29,546 Table 9 – Ethnic group by religion at 2001 (all) White Black Asian Other Religion 2001 Christian Buddhist Hindu Jewish Muslim Sikh Other No religion Not stated 372,022 8,727 1,857 1,860 545 28 176 737 67 33 6,097 55 2,549 12 32 39 2,077 1,166 13,494 849 72 4 4,495 43 1,149 66 239 47 72,119 1,469 773 1,734 36,372 1,416 1,440 520 Total 384,466 1,486 Source: ONS Longitudinal Study 6,252 2,632 17,586 4,614 1,501 76,095 39,748 Imputation in the 1991 Census • Edit & Imputation method • Edit matrices • Hotdeck method • Not flagged in the LS • Data imputed for wholly absent households “Editing and imputing data for the 1991 Census” I.Mills & A.Teague Pop Trends Summer 1991pp30-37 Imputation in the 2001 Census • Edit & Donor Imputation System (EDIS) • Edit • Imputation • One Number Census “Census 2001 Review & Evaluation Edit and Imputation: Evaluation Report” 2003 Table 10 – Imputation rates in the 2001 Census and the LS 2001 census LS Population Imputation rate Population Imputation 000's % % 49,359 0.56 539,879 49,359 0.44 539,879 49,359 0.32 539,879 49,359 1.30 539,879 48,848 1.70 538,026 48,848 2.91 538,026 48,848 1.09 538,026 48,848 1.42 538,026 48,848 3.92 538,026 48,848 4.53 538,026 35,367 391,841 35,367 391,841 35,367 6.09 391,841 35,367 3.69 391,841 33,686 6.14 369,784 33,686 9.15 369,784 33,686 3.34 369,784 21,741 3.48 244,104 29,335 13.81 369,784 369,784 28,065 4.73 304,380 Age Sex Marital status Student flag Country of birth Ethnic group Health Carer Long-term illness Address one year ago Educational qualifications Professional qualifications Highest qualification Activity last week Employment status Workplace size Supervisor Occupation - currently working Occupation - ever worked Economic activity Relationship to Person 1 Source: 2001 Census data from "Census 2001 Review & Evaluation - Edit and imputation: Evaluation Report" ONS Longitudinal Study rate 0.20 0.40 0.24 1.24 1.65 2.80 1.02 1.34 3.83 4.45 6.05 6.05 6.05 3.63 5.72 8.85 3.25 3.29 4.33 9.41 4.62 Table 10A – Imputation rates for Ethnic groups ALL PEOPLE WHITE White British White Irish Other White MIXED White & Black Caribbean White & Black African White & Asian Other Mixed BLACK OR BLACK BRITISH Black Caribbean Black African Other Black ASIAN OR ASIAN BRITISH Indian Pakistani Bangladeshi Other Asian CHINESE OR OTHER ETHNIC GROUP Chinese Other Ethnic group Source: ONS Longitudinal Study % 2.8 2.6 2.9 4.0 5.4 3.0 3.5 3.6 4.4 5.2 4.0 3.0 4.8 4.1 4.1 2.5 2.9 Table 10B - Relationships IMPUTATION RATES RELATIONSHIPS Person relationship Family relationship Household relationship Source: ONS Longitudinal Study N 529,874 529,874 529,874 % 8.47 10.37 11.39 Table 11 – Imputation rates for a derived variable All people aged 16 to 74 ever worked SOCIAL CLASS Variable Imputation rate % SOC2000 OCCP0 4.33 Employment status EMTP0 5.72 Supervisor status SUMP0 3.25 Size of establishment SIZP0 8.85 Social class Source: ONS Longitudinal Study 11.00 Imputation and analysis • Analysis with imputed values • Analysis without imputed values in the dataset • Analysis without imputed values in the variables of interest Table 12 – The effects of imputation Ethnic group by marital status (16-74 year olds, ever worked) Marital status (16-74 year olds, ever worked) Single Married/ Sep/Wdwd/ Remarried Divorced INDIAN With imputed values LESS 1659 54 6426 162 725 25 1605 187 6264 692 700 101 1418 5572 599 1521 61 1277 33 709 42 Without imputed values 1460 LESS status 232 Without imputed values 1228 for economic activity Source: ONS Longitudinal Study 1244 667 189 1055 131 536 Without imputed values LESS status Without imputed values for economic activity BLACK CARIBBEAN With imputed values LESS Change in ethnic group • Consideration of change in ethnic group is important for understanding the composition of groups over time • Almost 1 in 4 people recorded as Black African or Black Caribbean in the 1991 Census reported a different ethnicity in 2001, compared with less than 1 in 10 South Asian and Chinese people, and less than 1 in 200 White people • Change in individuals’ reported ethnic group between the 1991 and 2001 Censuses reflected the introduction of new response categories in 2001, especially for mixed ethnicities Per cent of people in selected 1991 ethnic groups reporting the same group in 2001 100 99.5 91.0 80 77.2 60 Per cent 40 20 8.3 0 White Indian Black Caribbean 1991 ethnic group Black Other 2001 ethnicity reported by people in the 1991 Other Black group Other Mixed 5% Other ethnic groups 7% White 15% Mixed White & Asian 2% Mixed White & Black African 6% Black Caribbean 21% Mixed White & Black Caribbean 33% Black African 3% Other Black 8% Loss to follow-up • Loss between two censuses without valid exit (death or embarkation) • Caused by: – Unobserved embarkation – Census underenumeration – Linkage failure (mismatched records) • Higher rates of loss to follow-up among minority ethnic groups – 16 per cent of 1991 Black Africans expected in 2001 were not found Follow-up of the 1991 traced LS sample by ethnic group White 78.7 Black Caribbean 11.7 65.0 Black African 6.2 59.3 28.9 3.5 Black Other 37.2 71.7 Indian 1.6 77.3 Pakistani 3.1 69.4 Chinese Other 10 20 Present in 01 30 28.4 6.1 30.2 71.4 0 28.5 4.3 63.7 3.1 40 50 60 Died or embarked 91-01 18.6 25.9 2.1 67.3 Other Asian 26.8 4.1 71.0 Bangladeshi 9.7 70 25.4 80 Not present in 01 90 100 Summary • Small sample sizes may constrain ethnic group analysis • Need to consider how to treat 2001 item imputation • Change in ethnic group between 1991 and 2001 should also be considered • Loss to follow-up is higher among minority ethnic groups – Potential for bias – Impact on sample sizes when measuring transitions