Supplementary questionnaire on the 2011 Population and Housing Census FRANCE

Transcription

1 Supplementary questionnaire on the 2011 Population and Housing Census FRANCE

2 Supplementary questionnaire on the 2011 Population and Housing Census Fields marked with are mandatory. INTRODUCTION As agreed during the last meeting of the "Working Group on Population and Housing Censuses" (4-5 June 2013), the transmition of the metadata on the 2011 census is split in two flows. In addition to the metadata provided using the ESS-MH (ESS Metadata Handler), Member States are asked to provide the set of additional information included in this questionnaire. The reason why these questions were not included in the ESS-MH is that they do not naturally fit into the ESMS (Euro-SDMX Metadata Structure), i.e. the standard structure for metadata transmission in the ESS. 0. Contact information Please insert the contact details of the person filling in this questionnaire Name / Surname Usage of estimation methods According to Annex II to Reg. (EU) No 519/2010, Member States shall report, for each census topic, on the methodology used to estimate data for that topic. For convenience, this question is applicable only to those topics for which estimation methods were used Were estimation methods used to estimate any topic in 2011 census? Yes No

3 If "Yes", please indicate for which topics estimation methods were used and provide a short description of the estimation methodology Adjustments are automatic and centralised for each entry batch (each year, all questionnaires are distributed between twenty batches of about 475,000 individual bulletins). They consist of two stages : first, total nonresponse adjustment (nonrespondant housing forms indicated as "FLNE"), then partial nonresponse adjustment (main use and complementary use variables). 1- Adjusting for nonrespondant housing forms (FLNE):To adjust fortotal nonresponse, the first step is to determine if the dwelling is a main residence and if so, the supposed number of occupants.in practice, a distinction is made between two types of FLNE: an FLNE is said to be "completed" if the enumerator has indicated the supposed number of occupants in the dwelling ; an FLNE is said to be "non-completed" if the enumerator did not know the supposed number of occupants in the dwelling. Completed FLNE are all considered to be main residences. Non-completed FLNE are considered to be "main residences", in a proportion identical to that observed for all completed questionnaires in the municipality. For non-completed FLNE considered as main residences, a supposed number of occupants is assigned to the dwelling based on the national distribution of persons per completed FLNE according to 4 categories (small metropolitan municipalities, small overseas municipalities, large metropolitan municipalitiesand large overseas municipalities). For all non-responding main residences are generated as many individual bulletins as necessary. The responses to these questionnaires are then assigned, as in the case of partial non-response. 2- Adjusting for partial nonresponse (main and additional use): Partial nonresponse is corrected by a sequential hot-deck imputation, batch by batch and variable by variable in accordance with a pre-established order of importance (defined in the document available on the site insee.fr at the following address in onnees_rp.htm).processing variables in logical order ensures the consistency of the answers for each person and the consistency of the characteristics shared by occupants of the same dwelling. It also facilitates detection and correction of inconsistencies in completed questionnaires. The retained technique assigns to any non-respondent the response of the closest individual in the entry batch sorted by census identifier: municipality, collection zone, address rank, dwelling rank, individual rank. If no donor is found, a deterministic (pre-established in function of the values encountered in the course of earlier collections) assignment is applied. All variables for all recorded observations are thus adjusted and finally filled in. Further information by topic is available in the metadata (item 14 about accuracy). 2. Usage of register-based data sources for the 2011 census This section of the questionnaire applies only to those Member States that used register-based data sources to produce the 2011 census data (possibly in combination with other data sources). Provisions concerning this part of the questionnaire are contained in Annexes I and II of Reg. (EU) No 1151/2010.

4 2.1. Which register-based data sources were used for which topics? (Please tick the appropriate boxes) Source / Topic Population register Social security/tax register(s) Employment/ occupation/ unemployment register(s) Education register(s) Business register(s) Building/ dwelling/ housing register(s) Other administrative register(s) Sex Age Current Activity Status Place of work Size of the locality Marital status Occupation Status in employment Educational attainment Place of birth

5 Country of citizenship Year of the arrival in the country Residence one year before Household status Family status Housing arrangements

6 2.2. Are there known problems of specific population sub-groups likely to be over or under covered by the registers? Yes No 2.3. Are there known reasons for people delaying registration/deregistration? Yes No 2.4. Has the extent of any problems in the completeness and timeliness of registration/deregistration been evaluated? Yes No 2.5. Were individual records matched across different registers? Yes No 3. Usage of sample surveys for the 2011 census According to item of Annex II of Reg. (EU) 1151/2010 some additional metadata are requested in case that Member States use sample surveys to derive part of the data of the 2011 census.

7 3.1. For which topics in the 2011 census were sample surveys used to collect statistical information? Topic Sex Age Current activity status Place of work Marital status Occupation Industry Status in employment Educational attainment Place of birth Country of citizenship Year of arrival in the country Residence one year before Household status Family status Housing arrangements

8 3.2. Please provide a short description of the sampling design used in these survey-based data sources The French census gathers information over a rotating five year cycle and produces results each year relative to the median year of the cycle. Thus, each year Y, the data is based on collections from the years Y-4 to Y and expressed in reference to the year Y-2.Given the large number and very great variety of French municipalities(almost 37,000, a quarter of which have a population of fewer than 200 inhabitants) a special collection system was implemented: Municipalities of fewer than 10,000 inhabitants (representing more than 36,000 municipalities) were divided into five groups balanced out at a regional level in accordance with ten criteria dating from the 1999 census and characterising the demographics of the municipality (population, sex, age group) and the dwelling stock (number of dwellings, number of main residences). In these municipalities all inhabitants and dwellings are surveyed exhaustively on the same date, whether main residences, communities (retirement homes, boarding schools, penitentiary establishments, etc.), mobile residences or homeless persons. In municipalities of more than 10,000 inhabitants in Metropolitan France (representing about 900 municipalities), a sampling system is used. The samples of addresses come from a survey address base extracted from the invoentory of adresses (RIL). This list describes all the addresses in the municipality. It is updated each year and is reviewed in collaboration with the municipalities. In order to take differences between addresses into account, they are divided up in the three following categories : large size addresses (at least 60 dwellings for the main criterion), new addresses and other addresses. The addresses in each of these categories are divided into five balanced groups according to the 1999 census criteria, whether demographic or linked to the number of constructions. Each year the annual census survey examines one rotation group : the large-size addresses and the new addresses in this group are exhaustively surveyed and known small addresses are selected so that the total sample (including new addresses and large-size addresses) represents about 40% of the dwellings in the rotating group i.e. 8% of the dwellings in the municipality. For other populations (living in communities, in mobile dwellings or homeless persons) the procedure is the same as in municipalities of fewer than 10,000 inhabitants. In municipalities of more than 10,000 inhabitants in overseas departments (Guadeloupe, Martinique, French Guiana, Reunion Island), in the absence of a RIL, a sampling system in two phases is adopted : the blocks of houses in each municipality are first divided into five balanced groups (with the same criteria as the Population Census 1999 in Metropolitan France: age groups, sex, dwellings). Then, each year a "cartographic survey" is applied to a group in order to localise all the constructions and count the dwellings therein. In the group thus constituted, a sample of addresses is taken with a survey rate of 40% of dwellings. The population of households in these municipalities is determined by accumulating the 5 data collections with their sampling weightings. For the other populations (living in communities, in mobile dwellings or homeless persons) the procedure is the same as for municipalities of fewer than 10,000 inhabitants. For further information, see Insee Methods "Understanding the population census", May 2005, Alain Godinot (available on the site insee.fr at this address in 2014:

9 3.3. Please provide an assesment of possible biases in the estimation due to model assumption errors To establish the results of the census for the desired reference date, all observations in the census are weighted each year. The method of estimation naturally differs according to the size of municipalities, in accordance with the sampling design : - For municipalities of fewer than 10,000 inhabitants, the method for calculating the weighting depends on the collection date and the population category, but all the municipalities are dealt with in a strictly comparable manner over the five-year period. In practice, the data collected for households listed in the municipalities surveyed in the reference year is preserved. For those listed before the median year, the household data is extrapolated by continuing the observed trend and making use of the number of dwellings provided by the local taxes (Taxe d habitation). Finally, for municipalities surveyed after the median year, the household data is interpolated with the household population from the previous Population Census. The same methods for interpolating, extrapolating or maintaining the collection are applied to populations living in communities, also taking into account the opening and closing dates of such structures. Lastly, the data collected for populations that are mobile by nature, those in mobile dwellings, and homeless persons, is maintained constant for 5 years. - For households residing in municipalities of 10,000 inhabitants or more in Metropolitan France, we first calculate the average number of persons per ordinary dwelling based on five successive collections (the estimated total household population is divided by the estimated number of ordinary dwellings, estimations based on the sampling weights). Then the household population for the municipality is calculated by multiplying the average number of persons per ordinary dwelling by the total number of ordinary dwellings (obtained from the RIL). This adjustment is done at a sub-municipal level (for zones of about 2,000 inhabitants, called IRIS districts). For the other populations (living in communities, in mobile dwellings or homeless persons), we proceed as with municipalities of fewer than 10,000 inhabitants. - To establish the results of the census for the desired reference date for municipalities in overseas departments (Guadeloupe, Martinique, French Guiana, Reunion Island) of 10,000 inhabitants or more, we calculate the household population for these municipalities simply by accumulating the 5 collections with their respective sampling weights. For the other populations (living in communities, in mobile dwellings or homeless persons), the procedure is the same as for municipalities of fewer than 10,000 inhabitants. For further information, see Insee Methods "Understanding the population census", May 2005, Alain Godinot (available on the site insee.fr at this address in 2014:

10 3.4. Please provide a short description of the algorithm used to calculate the standard error A SAS macro was developed in 2009 to evaluate the accuracy of the census results at a sub-municipal and a municipal level. It is based on the accuracy formulas developed by J.-C. Deville and Y. Tillé (2000) for estimators from balanced samples. This macro takes account of the balancing of addresses in 5 rolling groups and the adjustment with the IRIS using the residual method. However, it ignores the total and partial non-responses adjustments, as well as the temporal dimension (spreading the collections used over 5 years). 4. Assessment of the data sources used for the 2011 census According to item 2.4 of Annex I to Reg. (EU) 1151/2010, Member States shall report upon the extent to which the data sources used for the 2011 census meet the "essential features" as listed in Art. 4(4) of Reg. (EC) 763/2008 and defined in Article 2(2-6) of Reg. (EU) 1151/ To what extent do the data sources used for the 2011 census meet the "essential feature" of: A - Individual enumeration? Completely To a large extent Partially Poorly Not at all B - Simultaneity? Completely To a large extent Partially Poorly Not at all

11 C - Universality within the defined territory? Completely To a large extent Partially Poorly Not at all D - Availability of small-area data? Completely To a large extent Partially Poorly Not at all E - Defined periodicity? Completely To a large extent Partially Poorly Not at all 5. Transmition of data concerning primary and secondary homeless persons According to Reg. (EU) 519/2010 "data on the total population shall include the number of all primary homeless persons (persons living in the streets without shelter) and secondary homeless persons (persons moving frequently between temporary accommodation). However, Member States are free not to include the number of homeless persons in their data on the total population, or to include the number on the homeless but not to break the data on the homeless down by any breakdown or category (figure included only in the total and/or categorised under Not stated ). If Member States do not include the number of homeless persons in their data on the total population, they shall provide the Commission with the best available estimate for the number of all primary and the number of all secondary homeless persons in the whole Member State." This part of the questionnaire is designed to respond to these requirements and, for sake of clarity, reporting on primary and secondary homelessness is split into two separate sections. 5A) Primary homeless persons

12 5A.1. Which of the following sentences describes best the level of inclusion of primary homeless persons in the data transmitted for the 2011 census? Census data include primary homeless persons with detailed information for all census topics Census data include primary homeless persons with detailed information only for limited set of census topics Census data include primary homeless persons with no detailed information on any census topic Census data do not include primary homeless persons 5B) Secondary homeless persons 5B.1. Which of the following sentences describes best the level of inclusion of secondary homeless persons in the data transmitted for the 2011 census? Census data include secondary homeless persons with detailed information for all census topics Census data include secondary homeless persons with detailed information only for limited set of census topics Census data include secondary homeless persons with no detailed information on any census topic Census data do not include secondary homeless persons THANK YOU FOR YOUR PARTICIPATION!