Anders Wallgren and Britt Wallgren, Statistics Sweden.

• Preface xiChapter 1  Register Surveys – An Introduction 11.1 The purpose of the book 11.2 The need for a new theory and new methods 31.3 Four ways of using administrative registers 51.4 Preconditions for register-based statistics 61.4.1 Reliable administrative systems 71.4.2 Legal base and public approval 81.5 Basic concepts and terms 101.5.1 What is a statistical survey? 101.5.2 What is a register? 111.5.3 What is a register survey? 131.5.4 The Income and Taxation Register 141.5.5 The Quarterly and Annual Pay Registers 161.6 Comparing sample surveys and register surveys 201.7 Conclusions 23Chapter 2  The Nature of Administrative Data 252.1 Different kinds of administrative data 252.2 How are data recorded? 262.3 Administrative and statistical information systems 272.4 Measurement errors in statistical and administrative data 292.5 Why use administrative data for statistics? 302.6 Comparing sample survey and administrative data 322.6.1 A questionnaire to persons compared with register data 322.6.2 An enterprise questionnaire compared with register data 342.7 Conclusions 36Chapter 3 Protection of Privacy and Confidentiality 373.1 Internal security 383.1.1 No text in output databases! 383.1.2 Existence of identity numbers 393.2 Disclosure risks – tables 403.2.1 Rules for tables with counts, totals and mean values 413.2.2 The threshold rule – analyse complete tables! 433.2.3 Frequency tables are often misunderstood 443.2.4 Combining tables can cause disclosure 453.3 Disclosure risks – micro data 453.4 Conclusions 46Chapter 4  The Register System 474.1 A register model based on object types and relations 474.1.1 The register system and protection of privacy 534.1.2 The register system and data warehousing 534.2 Organising the work with the system 544.3 The populations in the system 564.3.1 How to produce consistent register-based statistics 574.3.2 Registers and time 584.3.3 Populations, variables and time 594.4 The variables in the system 604.4.1 Standardised variables in the register system 604.4.2 Derived variables 624.4.3 Variables with different origins 634.4.4 Variables with different functions in the system 644.5 Using the system for micro integration 654.6 Three kinds of registers with different roles 704.7 Register systems and register surveys within enterprises 724.8 Conclusions 74Chapter 5  The Base Registers in the System 775.1 Characteristics of a base register 775.2 Requirements for base registers 785.2.1 Defining and deriving statistical units 785.2.2 Objects and identities – requirements for a base register 805.2.3 Coverage and spanning variables in base registers 815.3 The Population Register 835.4 The Business Register 885.5 The Real Estate Register  935.6 The Activity Register 945.7 Everyone should support the base registers! 985.8 Conclusions 101Chapter 6  How to Create a Register – Matching and Combining Sources 1036.1 Preconditions in different countries 1036.2 Matching methods and problems 1056.2.1 Deterministic record linkage 1056.2.2 Probabilistic record linkage 1066.2.3 Four causes of matching errors 1126.3 Matching sources with different object types 1146.4 Conclusions 120Chapter 7  How to Create a Register – The Population 1217.1 How should register surveys be structured? 1217.2 Register survey design 1257.2.1 Determining the research objectives 1257.2.2 Making an inventory of different sources 1287.2.3 Analysing the usability of administrative sources 1287.3 Defining a register’s object set 1317.3.1 Defining a population 1317.3.2 Can you alter data from the National Tax Agency? 1347.3.3 Defining a population –  primary registers 1357.3.4 Defining a population –  integrated registers 1367.3.5 Defining a calendar year population 1377.3.6 Defining a population – frame or register population? 1387.3.7 Base registers should be used when defining populations 1417.4 Defining the statistical units 1427.4.1 Units and identities when creating primary registers 1437.4.2 Using administrative objects instead of statistical units 1447.5 Creating longitudinal registers – the population 1457.6 Conclusions 146Chapter 8  How to Create a Register – The Variables 1478.1 The variables in the register 1478.1.1 Variable definitions 1488.1.2 Variables in statistical science 1498.1.3 Variables in informatics 1508.1.4 Creating register variables – check list 1518.2 Forming derived variables using models 1518.2.1 Exact calculation of values using a rule 1528.2.2 Estimating values with a rule 1538.2.3 Estimating values with a causal model 1548.2.4 Derived variables and imputed variable values 1578.2.5 Creating variables by coding 1588.3 Activity data 1598.3.1 Activity statistics 1608.3.2 Activity data aggregated for enterprises and organisations 1618.3.3 Activity data aggregated for persons – multi-valued variables 1618.4 Creating longitudinal registers – the variables 1658.5 Conclusions 169Chapter 9  How to Create a Register – Editing 1719.1 Editing register data 1719.1.1 Editing one administrative register 1739.1.2 Consistency editing – is the population correct? 1759.1.3 Consistency editing – are the units correct? 1789.1.4 Consistency editing – are the variables correct? 1809.2 Case studies – editing register data 1819.2.1 Editing work within the Income and Taxation Register 1819.2.2 Editing work with the Income Statement Register 1839.2.3 What more can be learned from these examples? 1849.3 Editing, quality assurance and survey design 1859.3.1 Survey design in a register-based production system 1859.3.2 Quality assessment in a register-based production system 1869.3.3 Total survey error in a register-based production system 1919.4 Conclusions 192Chapter 10 Metadata 19310.1 Primary registers – the need for metadata 19310.1.1 Documentation of administrative sources 19410.1.2 Documentation of sources within the system 19510.1.3 Documentation of a new register 19510.2 Changes over time – the need for metadata 19510.3 Integrated registers – the need for metadata 19610.4 Classification and definitions database 19710.5 The need for metadata for registers 19810.6 Conclusions 200Chapter 11 Estimation Methods – Introduction 20111.1 Estimation in sample surveys and register surveys 20211.2 Estimation methods for register surveys that use weights 20311.3 Calibration of weights in register surveys 20411.4 Using weights for estimation 20711.5 Conclusions 208Chapter 12 Estimation Methods – Missing Values 20912.1 Make no adjustments, publish ‘value unknown’ 21012.2 Adjustment for missing values using weights 21412.3 Adjustment for missing values by imputation 21512.4 Missing values in a system of registers 21812.5 Conclusions 220Chapter 13 Estimation Methods – Coverage Problems 22113.1 Reducing overcoverage and undercoverage 22113.1.1 Coverage problems in the Population Register 22113.1.2 Coverage problems in the Business Register 22213.2 Estimation methods to correct for overcoverage 22413.3 Undercoverage in the administrative system 22613.4 Conclusions 228Chapter 14 Estimation Methods – Multi-valued Variables 22914.1 Multi-valued variables 22914.2 Estimation methods 23214.2.1 Occupation in the Activity and Occupation Registers 23214.2.2 Industrial classification in the Business Register 23614.2.3 Importing many multi-valued variables 23814.2.4 Consistency between estimates from different registers 24214.2.5 Multi-valued variables – what is done in practice? 24514.2.6 Additional estimation methods 24714.3 Application of the method 25114.4 Linking of time series using combination objects 25414.4.1 Linking time series 25414.4.2 Changed industrial classification in the Business Register 25614.5 Conclusions 258Chapter 15 Theory and Quality of Register-based Statistics 25915.1 Is there a theory for register surveys? 25915.1.1 Statistical inference at a national statistical office 26015.1.2 Theory-based methods or ad hoc methods 26215.1.3 The survey approach and the systems approach 26315.2 Measuring quality – why and how? 26715.3 Analysing administrative sources – input data quality 27115.4 Output data quality 27815.5 The integration process – integration errors 27915.5.1 Creating register populations – coverage errors 28015.5.2 Creating statistical units –errors in units 28215.5.3 Creating statistical variables – errors in  variables 28315.6 Random variation in register data 28815.7 The register system and data warehousing 29115.8 Conclusions 295Chapter 16 Conclusions 297References 301Index 305