180 likes | 308 Views
2007 Census Test Design. Dr. Jaime Collins Field Design, 2011 Census. Overview. Research Objectives General details of the design Local Authority (LA) selection criteria ED (Enumeration District) selection per LA ED selection by Enumeration Targeting Categorisation (ETC)
E N D
2007 Census Test Design Dr. Jaime Collins Field Design, 2011 Census
Overview • Research Objectives • General details of the design • Local Authority (LA) selection criteria • ED (Enumeration District) selection per LA • ED selection by Enumeration Targeting Categorisation (ETC) • Census Test Evaluation Survey (CTES) • Questions
Research objectives • 3 key objectives: • 1. Delivery method: post-out vs. hand-delivery • 2. Income: Income questions vs. no income • questions • 3. Outsourcing field staff recruitment, training • and pay
General details of the design 1 Date – Sunday 13 May 2007 *Reduced Sample – just over 100,000 households within 515 EDs within 5 LAs. 3 delivery attempts (hand-delivery only) and 3 follow-up attempts and reminder card 4 Treatment groups assigned at ED level – delivery method and income factors crossed.
General details of the design 2 Balancing • EDs in each treatment group balanced within each LA. • EDs and households balanced in each treatment group and ETC stratum, • i.e.5,000 households for each treatment group in each ETC strata: • (5,000 x 4) x 5 = 100,000 Power - Delivery method 6.7% within ETC stratum - Income 3% overall difference
LA sample selection • LAs selected: • Bath and North-East Somerset, Camden, • Carmarthenshire, Liverpool and Stoke on Trent • Key selection criteria: • a ‘representative’ sample of LAs found within • England and Wales; • one LA in Wales; • at least one LA that is enthusiastic for the Test; • at least one LA with rapid population movement; and • at least one LA with rapid development.
ED selection per LA Table 1: Distribution of EDs per LA
ED sample selection ED selection • 499 (97%) EDs randomly selected • 16 (3%) EDs purposively selected to take into account specific operational and/or practical issues *Original sample - 564 EDs. 48 randomly selected EDs dropped following Address Checking (8 from ETCs1-3, 12 from ETCs 4-5). 1 further ED dropped during data collection.
ED selection by ETC Table 2: Distribution of EDs by ETC
The Census Test Evaluation Survey 1 CAPI Survey carried out in June-July 2007 3 key objectives: • Measure coverage • Understanding/interpretation of new questions • Reasons for non-response Sample: • Sample of addresses within 87 EDs • 25/05/07 early respondent sample • 26/06/07 late respondent and non-respondent sample
The Census Test Evaluation Survey 2 Table 3: CTES response rates by 2007 Test response
The Census Test Evaluation Survey 3 CTES 2007 Test respondent questions: • CoverageCaptured 2007 Test data preloaded. Established usual residents/visitors in CTES and probed differences • Questionnaire design and content • Re-asked questions(language, qualifications and income) • Delivery (appearance and condition of questionnaire package) • Publicity(Advance card, information leaflet, follow-up card) • Future response method(internet) CTES 2007 Test non-respondent questions: (Reasons for non-response, non-respondent characteristics, internet, advance card).
Thank you Any questions?
Appendix 1 Table A1: EDs by ETC and LA
Appendix 2 Table A2: Removed EDs by ETC and LA
Appendix 3 Table A3: CTES 2007 Test respondents by ETC and LA
Appendix 4 Table A4: CTES 2007 Test non-respondents by ETC and LA
Appendix 5 Table A5: Variables used in ETC calculation