ANNEX IISupplementary provisions
ANNEX II Quality-related data and metadata The quality-related data and metadata about the data sources and topics comprise the items listed below. 1. RELEVANCE 1.1. Adequacy of data sources Member States have to report on the adequacy of the data sources, in particular on the impact of any major deviation from the essential features of population and housing censuses and/or from the required definitions and concepts where this seriously impairs the adequate usage of the transmitted data. 1.2. Completeness The following data have to be provided for — all geographical areas at the following levels: national level, NUTS 1, NUTS 2, — all hypercubes ( 1 ) and all primary marginal distributions ( 1 ) : (1) number of all special cell values ‘not available’ (2) number of special cell values ‘not available’ flagged as ‘unreliable’ (3) number of special cell values ‘not available’ flagged as ‘confidential’ (4) number of numerical cell values flagged as ‘unreliable’ 2. ACCURACY The following information: — has to be provided for each data source (section 2.1.) and each topic (section 2.2.), referring to person counts ( 2 ) and — may be provided for data sources (section 2.1.) and topics (section 2.2.), referring to counts of statistical units other than persons (optional) 2.1. Data sources ( 3 ) The data as required under point 2.1.1. have to be provided for all geographical areas at the following levels: national level, NUTS 1, NUTS 2. The explanatory metadata as required under point 2.1.2. have to be provided for the national level. 2.1.1. Data (1) Census population: absolute value and percentage of the estimated target population; (2) Estimated target population ( 4 ) : absolute value; (3) Under-coverage (estimated): absolute value and percentage of the census population; (4) Over-coverage (estimated): absolute value and percentage of the census population; (5) Number of all record imputations ( 5 ) : absolute value and percentage of the census population; (6) Number of all record deletions ( 6 ) : absolute value and percentage of the census population; (7) Additionally, for samples: complementary set of statistical units ( 7 ) : absolute value; (8) Number of non-imputed records in the data source for statistical units belonging to the target population: absolute value ( 8 ) , percentage of the census population ( 8 ) , percentage of the estimated target population ( 9 ) , and percentage of all non-imputed records in the data source (before any record deletion) ( 10 ) ; (9) additionally, for questionnaire-based data in the data source: ( 11 ) unit no-information (before record imputation): absolute value and percentage of the census population. 2.1.2. Explanatory metadata The explanatory metadata contain descriptions of — the operation to assess under-coverage and over-coverage, including information on the quality of the estimates for under- and over-coverage, — any method used to impute or delete records for statistical units, — any method applied to weigh data records for statistical units, — additionally for questionnaire-based data in the data source: ( 11 ) any measures to identify and limit unit no-information or other measures to correct errors during the collection of data. 2.2. Topics The data required under point 2.2.1. have to be provided for all geographical areas at the following levels: national level, NUTS 1, NUTS 2. The explanatory metadata required under point 2.2.2. has to be provided for the national level. 2.2.1. Data (1) Census population ( 12 ) : absolute value; (2) Number of data records ( 13 ) that contain information on the topic: non-weighted ( 14 ) absolute value, non-weighted ( 14 ) percentage of the census population; (3) Number of imputed data records ( 13 ) , ( 15 ) that contain information on the topic: non-weighted ( 14 ) absolute value, non-weighted ( 14 ) percentage of the census population; (4) Item imputation ( 13 ) ( 15 ) for the topic: non-weighted ( 14 ) absolute value, non-weighted ( 14 ) percentage of the census population; (5) Item no-information ( 13 ) (before item-imputation) for the topic: non-weighted ( 14 ) absolute value, non-weighted ( 14 ) percentage of the census population; (6) Number of non-imputed observations on the topic: ( 13 ) , ( 16 ) non-weighted ( 14 ) absolute value, non-weighted ( 14 ) percentage of the census population; (7) Transmitted data ( 17 ) for the hypercube set out in the table in Annex III for the topic in question ( 18 ) : absolute value, percentage of the census population; (8) Number of non-imputed data records ( 13 ) that contain non-imputed information on the topic broken down according to the hypercube set out in the table in Annex III for the topic in question ( 18 ) : non-weighted ( 14 ) absolute value, non-weighted ( 14 ) percentage of the census population; (9) Additionally, for topics about which information has been collected by means of a sample: coefficient of variation ( 19 ) for the cells in the hypercube set out in Annex III for the topic in question ( 18 ) . 2.2.2. Explanatory metadata The explanatory metadata contain descriptions of the method used to treat item non-response for the topic in question. For topics about which information has been collected by means of a sample, the metadata also contain descriptions of — the sampling design, — possible biases in the estimation due to model assumption errors, — formulae and algorithms used to calculate the standard error. 3. TIMELINESS AND PUNCTUALITY The following information has to be provided for the national level: (1) Calendar date(s) of the transmission of data to the Commission (Eurostat), broken down by hypercubes ( 1 ) ; (2) Calendar date(s) of major revision(s) of the transmitted data, broken down by hypercubes ( 1 ) ; (3) Calendar date(s) of transmission of the metadata ( 20 ) . In the case of major revisions on 1 April 2014 or after, Member States have to report the respective calendar date(s) separately to the Commission (Eurostat) within one week after each major revision. 4. ACCESSIBILITY AND CLARITY (OPTIONAL) Member States may report on the conditions for access to the data and metadata they make available from their 2011 censuses of population and housing, including on those relating to media, support, documentation, pricing policies, and/or any restrictions. 5. COMPARABILITY For each topic, Member States have to report on any definition or practice in the Member State which could impair the EU-wide comparability of the data. 6. COHERENCE For each topic referring to person counts ( 2 ) , Member States have to provide the average absolute deviation ( 21 ) for the cell values in the hypercubes set out in Annex III ( 18 ) . ( 1 ) As listed in Annex I to Regulation (EU) No 519/2010. ( 2 ) Topics, or data sources for topics, for which the total shown in the table in Annex III is the total population. ( 3 ) The reporting on the data sources must be comprehensive and free of overlaps in the sense that it is possible to allocate each topic to exactly one data source about which information is provided in this section. If a record linkage has led to the creation of a new data source, Member States have to evaluate the new data source rather than the original data sources from which the new data source has been derived. ( 4 ) ((1) + (3) – (4)), referring to the data in point 2.1.1. of this Annex, given in absolute values. ( 5 ) Any record imputation increases the size of the census population. In a data source resulting from a record linkage, only the records that have been imputed into any of the original data sources, thereby increasing the size of the census population, have to be counted as imputed records in the new data source. If a data record is weighted in the process of generating the required statistical output for the target population with a weight w orig bigger than 1, it has to be counted as an imputed record with the weight w imputed = w orig – 1. The reference hypercube for the weights w orig is the one listed below the table in Annex III for the statistical units on which the data source reports. ( 6 ) Any record deletion decreases the size of the census population. In a data source resulting from a record linkage, only the records that have been deleted in any of the original data sources, thereby decreasing the size of the census population, have to be counted as deleted records in the new data source. If a data record is weighted in the process of generating the required statistical output for the target population with a weight w orig smaller than 1, it has to be counted as a deleted record with w deleted = 1 – w orig . The reference hypercube for the weights w orig is the one listed below the table in Annex III for the statistical units on which the data source reports. ( 7 ) If a data source comprises, as a matter of methodological principle, data records for only a sample of the statistical units in its estimated target population, the size of the complementary set of statistical units is calculated according to the sampling design. ( 8 ) ((1) – (4) – (5) – (7)), referring to the data in point 2.1.1. of this Annex, given in absolute values, respectively 100 * ((1) – (4) – (5) – (7)) / (1). ( 9 ) 100 * ((1) – (4) – (5) – (7)) / ((1) + (3) – (4)), referring to the data in point 2.1.1. of this Annex. ( 10 ) 100 * ((1) – (4) – (5) – (7)) / ((1) – (5) + (6) – (7)), referring to the data in point 2.1.1. of this Annex. ( 11 ) In a data source resulting from a record linkage of more than one questionnaire-based data sources, the information has to be provided for each original questionnaire-based data source. ( 12 ) As identified under point 2.1.1. (1) of this Annex for the data source from which census information on the topic is derived for the target population. ( 13 ) For the census population in the data source from which census information on the topic is derived. ( 14 ) If the data records are weighted in the process of generating the required statistical output for the topic in question, ‘weighted’ means that these weights are be applied to the data records for the count, ‘non-weighted’ means that these weights are not be applied to the data records for the count. The reference hypercubes for the weights are those listed for the topics in the table in Annex III. ( 15 ) An item imputation has no effect on the size of the census population. For a topic belonging to a data source resulting from a record linkage, any record that contains information on that topic as a result of a record imputation into any of the original data sources is counted as a record imputation if the imputation increases the size of the census population, and as an item imputation for that topic if the imputation does not increase the size of the census population. ( 16 ) ((2) – (3) – (4)), referring to the data in section 2.2.1. of this Annex. ( 17 ) The data transmitted on the basis of Regulation (EU) No 519/2010 in the hypercube listed for the respective topic in the table in Annex III. ( 18 ) The geographical area for which the information has to be provided is indicated in the table in Annex III. ( 19 ) Where a numerical cell value is smaller than 26, the coefficient of variation can be replaced by the special value ‘not available’. ( 20 ) As listed in Annex II to Regulation (EU) No 519/2010. ( 21 ) The arithmetic average of the absolute (positive) value of the difference between the numerical cell value and its arithmetic average, with the arithmetic averages being calculated for all hypercubes (as listed in Annex I to Regulation (EU) No 519/2010) in which the respective hypercube as set out in Annex III is contained.