1 / 39

Alternatives in Census Designs

This article discusses alternative census designs, such as traditional census, supplemented by annual household surveys, registered-based census, and the use of administrative sources and sample surveys. It explores the concerns with traditional census methods and the potential benefits of alternative approaches.

celenad
Download Presentation

Alternatives in Census Designs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Alternatives in Census Designs Fourth Taskforce Meeting on Population & Housing Censuses Cairo, 4-5 September 07 Neda Jafar ESCWA jafarn@un.org

  2. Background • Most countries conduct traditional census • Headcount at a national level • statistical portrait/profile (composition, distribution & socio-economic correlates) • Governments uses: taxation, military service, planning of schools & hospitals

  3. Concerns • Falling response rates • Respondent burden • Cost • Data quality • Desire for more accurate information in non-census years • Lack of timeliness of results • Under coverage

  4. Population Census Designs • Traditional census • Traditional census supplemented by annual household surveys • Registered-based census • Administrative sources and sample surveys – virtual census • Sample surveys – rotating census

  5. A. Traditional census • Used by vast majority of countries especially outside Western Europe • Complete enumeration of population • Periodically (every 5-10 years)

  6. Scope for variation • Enumeration: • by interviewers • self completion form (enumerators/ via postal service) • Location: • Count at location or at usual place of residence • Questions: • Number and type of questions asked may vary extensively • Use of short & long form

  7. Australia(population: 19 million) • Every 5 years • A drop-off/pick up methods without any use of the postal service • Level of undercount determined by PES • In 2001 census included questions on use of PCs and internet • 1st time respondents opt giving permission for releasing record in 99 years • In 2006 traditional census

  8. AustraliaThe Future • Two major changes planned • Fill in the census form via internet • Use post-back in buildings/apartment blocks (difficult to obtain a response thru pick-up)

  9. Canada(population: 30 million) • Usual residents, every 5 years • A drop-off/post-back methods since 1971 • Short form contains 7 questions on basic demographic information • Long form sent 1/5 households on parents’ birthplace, single-sex couples and detailed language question

  10. Canada • Gross under coverage rates have been steadily increasing reaching 3.3% in 1996 • In 2006, 2/3 of forms posted out + comprehensive address register • Possible to use the internet to complete the questionnaire

  11. B. Traditional census supplemented by annual hh surveys • A traditional census carried out 5/10 yrs • Traditional short form • USA “Re-engineered Pop and Housing Census • Prior census, large-scale annual surveys • Complemented by Community Survey Programme • Surveys since 2003 • Replace long form

  12. USAAmerican Community Survey • Collection on monthly samples • Each sample has 3-month collection cycle • Each collection cycle has 3 phases: • Questionnaire mailed • Use of computer-assisted interviews to follow up on non respondents • Personal interview to non-respondent

  13. USAThe Future • 2010 census will run on traditional lines • No Long form • Use adm records data for non response hh • ACS with annual sample size of 3m hh • 10years ¼ population cover (>long form) • Low level/annual pop reliable estimates for areas/pop >65,000 available 1st yr of the national survey

  14. C. Register-based census • Data from adm. registers combined at micro level • No questionnaires are sent • Covers all private hh and dwellings • In principal, can replace a census completely. • In practice, constrained by the quality & extent of data in registers

  15. Italy(population: 57 million) • Every 10 years since to 1861 • Municipal population registers, but each municipality is responsible for its own register and these vary highly in quality • Little incentive to remove people who have died or moved

  16. Italy • Significant difference between resident & registered population • Reasons include: • Students registered in one municipality but live n another • People temporary resident in one municipality due to work • People performing military service

  17. Italy • Traditionally census was used to update the registers • Registers are used to estimate undercount in census • In 1991 the raw register figures gave 5% undercount (attempts to remove superfluous register data reduced this figure to 1%)

  18. ItalyThe Future • ISTAT attempt to standardize population registers and other adm sources to conduct register-based censuses in the future

  19. Spain(population: 41 million) • 2001 census was the first in the world to allow all residents to complete their forms via internet • The existence of municipal population registers was central to census operation

  20. Spain • Paper questionnaires were pre-printed with information from the registers, to be corrected, if necessary • Identity code and password to enable access to the on-line version • However only 0.1% of households responded

  21. D. Adm. sources & Sample surveys • Some countries obtain census information by compiling data from: • different adm registers, and • results of sample surveys

  22. The Netherlands(population: 16 million) • 1981 traditional census approach abandoned • Stat Netherlands used adm registers and existing sample surveys • However, data at aggregate level i.e. not possible to integrate them fully with survey data • Legislation allowed administrative records to be used to their full potential for statistical purposes, to reduce costs & respondents burden

  23. The Netherlands • Dutch are expected to register with local authorities. • It contains name, address, and basic demographic data (sex, date of birth, and marital status) • Updated once a year on 1 January

  24. The NetherlandsSSB • In 1996, a Social Statistics Database (SSB) contains information at an individual level • Covers residents during year based on pooled population register created at the end of the year • Population register is the spine • Other adm records sources: employee insurance schemes, social security, water electricity, pension

  25. The NetherlandsSSB • The database enables the statistical office to come up with census related results at any chosen time. • File created in a micro-integration process from diff sources • Transformed and adjusted to conform with concepts and definitions set in the census programme • “Repeated weighting” used to generate numerically consistent table sets for data from different sources

  26. The NetherlandsSSB • Matched using variables: sex, date of birth & address • The matching process aims to match over 90% records with no > than 5 % false matches • No Coverage estimate due to lack of another measure of the population • In 2001 called “Virtual Census” based on combined info from diff sources using personal id codes

  27. Denmark (population: 5.4 million) • Central population register (CPR) based on municipal registers • Change in address is compulsory, with birth and death registration update CPR • Other adm registers linked to CPR by unique personal identifier • A register-based population census took place in 1976 and following the establishment of a register of buildings and dwellings, a register-based population and housing census in 1981 • Neither of these included any form of fieldwork

  28. Denmark • Statistics Act obliges public bodies to make available any information required by Statistics Denmark for statistical purposes • Consult Statistics Denmark when developing registers to ensure that they are appropriately organized for statistical use • Population statistics are produced every year from the CPR and other administrative sources • Population registers associated with the issuing of identity cards are held at municipality level

  29. E. Sample Surveys • Large-scale continuous annual surveys or rolling survey • Cover whole country over a period of time rather than on one particular day • Cover a region in its entirety or by taking a sample from every area • Merits • Cost effective • Builds national capacity of NSS • Quality of data on annual basis • Generate new up-to-date stat info for users • Annual estimates from very large sample • Annual time series for various adm domains

  30. France • France terms it a “rotating census” • Annual collection of data, during a five-year period • Based on sample surveys providing pop data for the adm districts and annual socio-demographic statistics • Adopted for 4 main reasons: • No pop registers • Several large-scale surveys with results that can be inferred at relatively low domains • While adm data may be available, the files can not be linked coz there are no personal id number system • Until 1999 no legal obligation to carry out a census and no clear legal framework for conducting traditional census

  31. Rotating Census – New model • Cover every household over a given period of time (5-10 years) • Proportion of population would not be counted at all in 5 years cycle • Census population <10,000 in communes enumerated once every 5 years • Larger communes a sample survey with 40% population will be enumerated i.e. any given commune will be sampled every year • Sampling frame drawn from building register • Building register was created initially from 1999 census • Updated continually from construction & demolition permits also other sources & surveys

  32. Conclusions • Only within Western Europe that a complete enumeration is not always attempted • The Scandinavian countries having good population registers or are about to be, the sole means of enumerating population • The Netherlands has a unique model that makes good use of all the sources available-population registers, other administrative sources and sample surveys

  33. Conclusions • France use a rolling model which is part census and part sample survey, and it depends on at least one good administrative sources (building registers) • Italy is looking to develop a more standardized system that can be used to conduct register-based censuses • USA is moving to ‘head count’ plus ‘survey’ model. USA does not have a compulsory population register

  34. Conclusions • No two countries are exactly alike even for those that still conduct a traditional census • Not every model would work for every country • Factors depends on • population size, • quality of administrative records, • population base issues, • public perception, • political climate, • funding, • legislative and constitutional issues • existence of a reliable population register

  35. Towards a census information-system • The ideal situation is not to one approach • Optimize use of available sources • To avoid faults and to take advantage of the merits of each. • Shift from a census system to a census information-system where the data are: - Rich - Detailed - Dynamic - Continuously updated - Monitors flows ofsequent stock of snapshots

More Related