PART 1 METHODOLOGICAL AND OPERATIONAL ASPECTS OF CENSUS TAKING
5. OPERATIONAL FRAMEWORK II: LEGISLATION, SECURITY,
CONTROL
Introduction
In most countries, the preparation and conduct of a census and/or the collection and compilation of statistical data from administrative sources requires a legal basis, regulating issues such as: the allocation of funds for the census operations; the obligation of citizens to provide census information; the relationships between the agency responsible for the census and other public administrations involved in the census operations; the uses and linkage of registers to produce census data or to support field operations; and data security and confidentiality.
The last of these elements is particularly important. The census collects information on each person and household in the country. In its uses it is not concerned with facts about individuals as such. Its purpose is to provide statistics about the community, and groups within the community, as a whole. The public, therefore, has a right to expect, and needs to be assured that, personal information provided in confidence will be respected. The confidentiality requirement, whether enshrined in legislation or not, encompasses the whole census operation, ranging from the security of the completed census questionnaires both in the field and during processing, to the protection of the information contained in the outputs and made publicly available.
This chapter reports on both the legislative framework underpinning the 2010 round censuses in the ECE region and the practices adopted by countries to ensure the security and confidentiality of the information collected9.
Legislation
In many countries, a specific census act or appropriate regulations are approved before each census, both to authorise the topic content and to deal with the issues mentioned above. In some countries, however, more general statistics legislation includes all the necessary provisions required for the conduct of a population census and/or the production and dissemination of statistical data thereby obviating the need for specific census legislation.
One of the main issues covered in census and statistics acts is data confidentiality. In an increasing number of countries, specific data-protection laws have been approved to regulate this field. In some cases, data-protection laws include all necessary provisions to cover the specific needs of censuses, including for instance the possible use of register data for censuses, or specific measures to be applied to census enumerators. In others cases, specific provisions on data confidentiality have to be included in the census acts, to take into account aspects, which are specific to the census.
Table 5.1 shows that regardless of the census methodology adopted, all countries that responded to the UNECE survey question (49) either reported or were known to have appropriate legislation in force to allow the collection of information necessary for the production of census statistics in the 2010 round.
9 The material in this chapter has been taken largely from reports of the UNECE survey prepared by Ian White (UK Office for National Statistics) on legislation and security and Eric Schulte Nordholt (Statistics Netherlands) on confidentiality and disclosure control, and presented to the Joint UNECE-Eurostat Work Session on Population and Housing Censuses in Geneva from 30 September to 3 October 2013 (http://www.unece.org/stats/documents/2013.10.census1.html)
48 Table 5.1
Census legislation in UNECE countries, by type of census
Countries EEA member
Type of legislation Data protection legislation
Status of legislation Census General
Statistics Permanent Unique/
revised Traditional census
Albania X X X
Armenia X X X
Azerbaijan X X X
Belarus X X X
Bosnia and
Herzegovina X X X
Bulgaria X X X
Canada X X X
Croatia X X X
Cyprus X X X
Czech Republic X X X
France X X X
Georgia X X X
Greece X X X
Hungary X X X
Ireland X X X
Italy X X X
Kazakhstan X X X
Kyrgyzstan X X X
Luxembourg X X X
Malta X X X
Montenegro X X X
Portugal X X X
Republic of Moldova X X X
Romania X X X
Russian Federation X X X
Serbia X X X
Slovakia X X X
Tajikistan X X X
United Kingdom X X X
United States X X X
(continues on next page)
49 Countries EEA member
Type of legislation Data protection legislation
Status of legislation Census General
Statistics Permanent Unique/
revised Register-based census
Austria X X X
Belgium X X X
Denmark X X - -
Finland X X X
Iceland X X X
Netherlands X X X
Norway X X X
Slovenia X X X
Sweden X X X
Combined census
Estonia X X X
Germany X X X
Israel X X X
Latvia X X X
Liechtenstein X X X
Lithuania X X X X
Poland X X X
Spain X X X
Switzerland X X X
Turkey X X X
Total countries:
Total UNECE 49 31 19 49 21 27
Total EEA 30 18 13 30 15 14
Total EU 27 17 11 27 13 13
Total traditional 30 23 7 30 9 21
Total register-based 9 1 8 9 8 0
Total combined 10 7 4 10 4 6
Legislation that was specific to the census was in force in 31 countries in total (63 per cent of the countries that responded) – with a similar proportion (60 per cent) among the 30 countries of the EEA. This includes the United Kingdom where specific census legislation exists in Scotland and Northern Ireland, but where in England and Wales several elements of its long-standing Census Act of 1920 have now been amended by the provisions of newly introduced statistics legislation (The Statistics and Registration Service Act 2007). Some 19 countries (39 per cent) had in place other, more general, legislation providing for the collection of population and other statistics, of which 13 were in the EEA (43 per cent of EEA countries). In Israel specific provisions relating to each census are added to its more general statistics legislation as and when they are required.
50
In comparing the type of current legislation in countries adopting different census methodologies it is perhaps not surprising to note that 23 out of 30 countries adopting a traditional approach (77 per cent) have legislation specific to the census (with 12 out of 14 in the EEA), whereas among the nine register-based countries all but one carry out their censuses under more general statistics legislation. Where countries adopt a combined methodology, incorporating elements of both traditional field enumeration and the use of administrative data sources, the split is similar to that for countries with traditional census: 7 countries out of 10 have legislations specific to the census, and in the other 3 countries the general statistics legislation covers the census.
In 21 of the 48 responding countries (44 per cent) the legislation governing their census is a permanently enshrined enactment, while in 27 countries (56 per cent) the relevant legislation either has to be amended or revised afresh for each successive census or was unique to the census in the 2010 round. The respective numbers among the 28 responding EEA member states were pretty evenly split at 15 and 14. Although the UK reported that its relevant primary (framework) legislation is permanent, it should be noted that secondary legislation, in the form of regulations, is required each time in order to implement particular elements of the census operation, covering such aspects as the duties of the field staff sand the inclusion of some topics on the census questionnaire. A similar legislative framework exists in Ireland.
Security, confidentiality and disclosure control
In all responding countries, there is legislation of some form in place (be it specific to statistical confidentiality or to data protection more generally) that protects the confidentiality of personal information collected for census purposes (Table 5.1). Moreover, all but one country (Georgia, where the census is planned to take place in 2014) reported having a formal policy and/or strategy for ensuring the security and confidentiality of such information – although in seven countries (three of which undertake solely register-based censuses) the details of the policy is not in the public domain.
In about half of the responding countries personal census information or microdata is made accessible to persons outside the NSI for the purpose of scientific or statistical research while it remains closed to public inspection. Different conditions and restrictions nevertheless apply to such access in different countries (see Chapter 6). In particular it will be noted from Table 5.2 that among countries adopting a traditional census a small majority offer no access to microdata, while eight out of the nine countries conducting a wholly register-based census do so. The split among the ten countries adopting a combined approach methodology is even.
Of those countries that make such personal census information or microdata accessible for scientific or statistical research, all countries imposed safeguards to protect the security and confidentiality of the data. As well as the imposition of appropriate IT safeguards and security procedures, such measures included ensuring that field staff and data processing staff were made fully aware of their legal obligations and the penalties for any infringement of confidentiality rules. In the United States for example, all people employed on the census were required to take an oath to protect personal data from disclosure and to undertake annual refresher training.
Table 5.2
Access of personal census information or microdata (number of countries) Accessibility to personal census information
or microdata
Total countries
Type of census
Traditional Register-based Combined
Yes 26 13 8 5
No 24 18 1 5
51 Furthermore, most countries take measures to protect the statistical confidentiality of published output from the census. Post-tabular methods were reported as being more commonly adopted than pre-tabular methods irrespective of census methodology or region (Table 5.3). It should be noted that a similar enquiry undertaken by Eurostat in 2012 into protection measures being adopted in EU member states showed a number of countries had not taken final decisions on the precise way of protecting their census outputs. In a follow-up survey by Eurostat in 2013 this was still the case.
Table 5.3
Measures to protect the statistical confidentiality (number of countries) Measures to protect the statistical confidentiality of
published output from the census Total countries
Region
EEA Non-EEA
Pre-tabular measures only 8 7 1
Post-tabular measures only 17 13 4
Both pre-tabular and post-tabular measures 21 9 12
No measures 2 0 2
Table 5.4 notes the specific measures that countries reported having planned to protect the statistical confidentiality of published output from the census. Restricting the number of output categories into which a variable may be classified in any table implies introduction of global recodes and is the most popular (post-tabular) method (36 countries adopted this process). Global recodes are very commonly applied to the age variable in aggregating single years of age into broader groups, but could be applied in principle to any census variable. Applying minimum population and/or household thresholds for outputs for small areas was also adopted by almost as many countries (33).
Among the methods used to modify the data, cell suppression is the most commonly adopted (by 18 countries). In contrast, small cell adjustments, which was a source of annoyance of users of the UK census in 2001, was adopted by only two countries (Malta and Poland).
The method of statistical disclosure control adopted by countries does not appear to be a correlated to any degree by either their census methodology or whether or not they are in the EEA.
Table 5.4
Statistical disclosure protection measures (number of countries) Measures planned to protect the statistical confidentiality of
published output from the census
Total countries
Region EEA Non-EEA Restricting the number of output categories into which a variable
may be classified in any table (such as aggregated age-groups) 36 23 13 Applying minimum population and/or household thresholds for
outputs for small areas 33 19 14
Modifying the data in one or more ways (indicate all that apply) 22 17 5
Cell suppression 18 12 6
Rounding 9 7 2
Record swapping 6 4 2
Over-imputation 2 0 2
Small cell adjustment 2 2 0
52
The UNECE survey on this topic concluded by asking countries to report whether or not they had commissioned an independent review or reviews of the measures taken to protect the physical security and/or statistical confidentiality of census information, and whether or not such reviews were or would be published. Most countries (30 out of 47) had not undertaken any such review.
53