October 28-30, 2009 UNECE Geneva Quality Assessment of 2008 Integrated Census - Israel Pnina ZADKA Central Bureau of Statistics Israel
24 October 2009UNECE Geneva Content of Presentation The integrated census in a nutshell Improved Population Register (IPR) Quality Geographic infrastructure quality Area sample survey data collection quality Over-count survey data collection quality Automated procedures quality
24 October 2009UNECE Geneva The integrated census in a nutshell Population register improved by administrative sources (IPR) Administrative families Two independent samples –Area sample to estimate under-count (CAPI) –IPR sample to estimate over-count (CATI) Editing and imputation Calculating weights Final demographic estimates
24 October 2009UNECE Geneva An accumulative register – nobody is deleted “Inactive” persons records –Deaths –Very old people “Active persons” “Suspected inactive” Border control files National Insurance Institute files Improved Register Quality
24 October 2009UNECE Geneva IPR Quality assessment – “active” persons 8 million “active” records in the population register 518,000 “active” persons, re-defined as “inactive” in the IPR based on external administrative sources 0.46% (33k) were miss-classified as “active” in IPR 0.04% (2.8k) were miss-classified as “inactive” in IPR
24 October 2009UNECE Geneva 80% of the active persons live in their IPR registered “statistical area” (census tract) 12% of the active persons live in their IPR locality in a different “statistical area” 8% of the active persons live in a different locality IPR Quality assessment -address
24 October 2009UNECE Geneva Geographic infrastructure Building Layer was frozen 18 months prior to under-count survey data collection process (updated up to one year lag) Improving orientation land marks were added to the computerized maps Pre-census enumeration updating of new residential buildings – 9032 (5%) buildings added
24 October 2009UNECE Geneva Under-count survey quality Coverage – 5% less dwellings than expected, according to the number of administrative households –83% of dwellings were interviewed 78% full questionnaires 5% partial questionnaires –4% refusals –9% no contact (un-occupied, closed, business)
24 October 2009UNECE Geneva Reliability 5% of the households were re-interviewed over the phone using a short questionnaire by a superior. Rare cases of misconduct (less than 0.5% in total) whenever possible they were re- interviewed on the full questionnaire Under-count survey quality
24 October 2009UNECE Geneva Consistency verification during the interview –“Warning” an alarm to the interviewer to verify the response –“Error!” notifying the interviewer of an incorrect response preventing the continuation of the interview –Incorrect control digit of Personal Identification Number (PIN) Not yet quantified Under-count survey on line assessment testing
24 October 2009UNECE Geneva On line auditing Daily auditing of the previous day data transmissions Set of pre-defined queries, updated along the field work All (100%) of the questionnaires were audited Feedback was sent to the regional staff each morning (before the interviewers start their next day of work)
24 October 2009UNECE Geneva Types of queries - examples Reduced household size Information not obtained from interviewee Skipping questions Missing critical information
24 October 2009UNECE Geneva On line validation Daily matching with IPR to validate identification parameters –PIN –Date of birth –Given name –Father’s given name
24 October 2009UNECE Geneva Over count survey -assessment Telephone interview procedure –Recording all interviews, listening to a sample of previous days interviews –Incognito on line listening to a sample of interviews –Miss conducts rare, repeated interviews for interviewers with miss conducts
24 October 2009UNECE Geneva 77.2% - response rate to telephone interview 5.1% - without telephones (no telephone registered on their name or a first kin such as parents, children, siblings) 6.2% - wrong numbers 11.2% - “no response” Over count survey -assessment
24 October 2009UNECE Geneva 82.5 % - Census tract (full address) 4.4% - Locality 13.1% - Unknown locality (deceased, abroad less than 12 months, locality’s name does not exist, un-known locality, refused to report) Over count survey –resolutions assessment
24 October 2009UNECE Geneva Over count survey secondary uses Assessment of Under-count field work –1% of persons were omitted in the under-count survey (entire household omitted or persons omitted from household questionnaire)
24 October 2009UNECE Geneva Timeliness Data collection over 6 months Automated procedures –Matching (97%) –Coding (75%) –Editing and imputation (100%)
24 October 2009UNECE Geneva Timeliness Demographic results for localities - two months after completing of the field operation Socio-demographic results ten months after the field operation
24 October 2009UNECE Geneva Thank You