Using Census Microdata for demographic research Sabine Springer, Christophe Bergouignan CENSUS project (INED, ODE, IEDUB) European Doctoral School of Demography 2009 May 19 CENSUS

Using Census Microdata for demographic research Census microdata samples databases. Census microdata what for ? IECM-IPUMS databases an easy way to get census microdata. Comparisons across time and space – International comparisons with IECM-IPUMS databases (the portugese example). Studying internal migrations with census microdata. – Pointing out interactions between factors of mobility and type of move, – The particular case of interactions between fertility and mobility.

Census microdata samples databases.

Census microdata what for ? Still few users – Availability and ethical issues (indirect identification, « sensitive » variables,…), recent answers of data suppliers, – Lack of variables and some kinds of inaccuracy (indirect estimates, answers relialability,…), – Researchers representations (« photograph of present », overestimation of innacuracy of indexes,…..). Census microdata samples / Census aggregate macrodata, less availability and more sampling errors but : – More cross-sectional options (complex indexes, statistical models including variance of indidividuals, micro- simulations), – Analysis involving household numbers or family ties (families or households new classification, measuring behaviour of individuals according to households characteristics, indirect estimates of demographic indexes,…). Census microdata samples / Survey micro-data, less variables and less accuracy for life-courses disaggregation, but : – More statistical significance/representativity : Leading to more cross-sectional options (age-detailed or cohort-detailed analysis useful for phenomena with high age or cohort concentration, e. g. students migrations), Leading to more spatial disaggregation (geographical interactions, local specificities). – Mostly a lower non-response rate (1% to 10% for censuses, 0% to 55% for surveys ; e. g. around 20% for EHF 1999 / around 2% for french 1999 census).

IECM-IPUMS databases an easy way to get census microdata IECM-IPUMS project : – Availability of census microdata, – Comparability of census microdata. IECM-IPUMS website : – Metadata, – Microdata to download. CENSUS

Comparison across time and space Classification variability among countries – Data processing to harmonize (family ties,…), – Data collection (students home and generally households « members » living outside the dwelling,…), – Enumeration forms (previous residence,…), – National statistical « custody » (french « PCS »,…), – Multidimensionnal meanings of variables (e. g. : educational attainment,….).

General problems in cross country comparison or in comparison in time Differences in the underlying concepts – meaning or interpretation Differences in the operational definitions of these concepts – meaning and interpretation Differences in the measure of the concept – choice of question and its wording, differences in the proposed answer categories Differences in the definition of the universe CENSUS

Example: Multigenerational households in southern Europe The tool Advantages Possibilities Limits Pitfalls Shortcomings CENSUS The Analysis Descriptive: Frequency Composition Function Comparison: Change over time Differences between countries

Multigenerational households in Europe Why census micro data? Why IECM data base? Are there alternatives? What are multigenerational households? How can we identify them? What do we want to know about them? CENSUS

Comparison of multigenerational households across time and space Portugal (1981, 1991, 2001), France, Italy Identification of Household Members Generation CENSUS The same for the three censuses? The same for the three countries?

Identification of household Household Types Private household Institutional household Collective household Homeless household Info can be found on the IECM website under : Meta data : Source: Variable: CENSUS

Fernando Casimiro, F. Casimiro, 2009, INE, Portugal

Definition of private household CENSUS Portugal (1981, 1991, 2001) Family Group of persons living at the same dwelling and that have kinship relations among them ("de jure" or "de facto"), regardless of occupying the whole or part of the housing unit. A person who lives alone in a separate housing unit or who occupies, as a logder, a separate room (or rooms) of a housing unit, but does not have kinship relations with the other occupants project.org/index.php?module=metadata&c=prt France (1990) Dwelling unit no definition given project.org/index.php?module=metadata& c=fra&y=1990 Italy (2001) Family The term household refers to a group of people, bound by marriage, kinship, affinity, adoption, guardianship or by emotional ties, who are partners and live in the same Municipality (even if still not registered in the Population Register residing in that Municipality). A household may also be composed of one individual only. shtml#it shtml#it Spain (1991) Housekeeping Group of people who, residing in the same dwelling, share expenses derived from the use of the dwelling and/or alimentation. Single person and multiperson households are to be considered. project.org/index.php?module=metadata& c=esp

Type of households Integrated gq - Group quarters status 00 Vacant 10 Households 20 Group quarters 29 1-person household created by splitting large unit GQ identifies households as vacant dwellings, group quarters, or private households. Group quarters -- collective dwellings -- are generally institutions and other group living arrangements such as rooming houses and boarding schools. Institutions typically retain persons under formal supervision or custody, such as correctional institutions, military barracks, asylums, or nursing homes. Some of the usual information for households is not available for group quarters and vacant units. action/variableDescription.do?mnemonic=GQ Not harmonized pt81a_dwtype1 Type of living quarter 1 Conventional dwelling 2 Non-conventional dwelling 3 Collective living quarter/Other rg/international- action/variableDescription.d o?mnemonic=PT81A038 pt81a_gq Household type 1Private Household 2Institutional Household 3Vacant household rg/international- action/variableDescription.d o?mnemonic=PT81A055X CENSUS Portugal 1981

Integrated - Group quarter status CENSUS

Not integrated CENSUS

Comparison CENSUS

Type of households Integrated gq - Group quarters status 10 Households 21 Institutions 22 Other group quarters Not harmonized FR82A022 FR82A022 (FR82A_DWTYPE) Label: Type of dwelling 0Collective dwellings 1Ordinary dwelling 2Assisted living for the elderly 3Independent rooms rented, sublet or lent to individuals 4Furnished room(s) in a hotel, a boarding house, a privately-run hotel, etc. 5Make-shift housing or temporary housing used for lodging FR90A047 FR90A047 (FR90A_CATPOPHH ) Label: Category of the population, household 01 Population counted as a household: counted in a dwelling 02 Population counted as a household: reintegrated into a dwelling; boarding student reintegrated 03 Population counted as a household: reintegrated into a dwelling; military reserve or military by career 11 Population of collective groups: worker living in a workers' housing development complex 12 Population of collective groups: student in a university dormitory or student home 13 Population of collective groups: person in a retirement home or nursing home 14 Population of collective groups: hospitalized person 15 Population of collective groups: member of a religious community 16 Population of collective groups: person in a youth hostel or a host family 17 Population of collective groups: another case of a person in a collective group 21 Population having mobile homes 22 Navy 31 Population of an establishment: boarding student 32 Population of an establishment: military reserve or military by career 33 Population of an establishment: prisoner 41 Population capable of being integrated but not integrated: boarding student capable of being reintegrated but not reintegrated 42 Population capable of being integrated but not integrated: military reserve or by career capable of being reintegrated but not reintegrated CENSUS France 1990

Identification of the members -De jure or de facto census concept -Resident status -Problematic groups: -Students -Working migrants -Living together apart -Visitor, migrants (emigrants, imigrants) -Homeless persons, persons without a fixed living quarter CENSUS

Identification of the generation – Relationship with household head Who is head of household? Every categories allows the identification of a generation ? => Otherwise : Age CENSUS

Conclusion + + possibility of comparison between countries that does not exist otherwise + integrated variables that allow the comparison between countries + explicit meta data information and census documentation + unharmonized data available CENSUS

Conclusion - -translation not always correct -integrated variables– false reliability -Problems with data quality -particular problems for specific samples CENSUS

Conclusion =  IPUMS data base can not be better than the census itself, but it can indicate problems!  There is no alternative for certain research subjects CENSUS