Q2014 european conference on quality in official statistics
        Special session: Consistency of Concepts and Applied Methods in Business Statistics
   Introduction
   In order to guarantee the quality of results of European statistics
   according Regulation (EC) No 223/2009 Article 12 seven criteria are to
   be met. 
   The first five of them refer to the quality of a statistical domain
   (often a single variable) seen in an isolated way. 
   The last two criteria ‘comparability’ and ‘coherence’ address the
   relationship between statistical projects; they deal with system
   wide aspects.
   In many cases the assessment of quality concentrates on the first
   five criteria and in particular on the criterion of accuracy.
                                                                         2
         Norbert Rainer, Josef Richter                                Vienna, June 2014                                                                                   2
             Q2014 european conference on quality in official statistics
        Special session: Consistency of Concepts and Applied Methods in Business Statistics
   Introduction
   Comparability and coherence and the system aspect of quality are of 
   special relevance in business statistics.  
   This aspect was taken up by the “ESSnet on Consistency”.
   Based on the “Consistency study” which identified the inconsistencies laid
   down in the legal framework of European business statistics, the 
         objective
   was 
   •     to prepare an inventory of the implementation of the concepts in the 
         Member States and 
   •     to derive appropriate proposals how inconsistencies can be 
         reduced.
                                                                         3
         Norbert Rainer, Josef Richter                                Vienna, June 2014                                                                                   3
             Q2014 european conference on quality in official statistics
        Special session: Consistency of Concepts and Applied Methods in Business Statistics
   Introduction
   Inconsistencies are caused by 
   •     inconsistent concepts
   •     different ways of implementing concepts
   •     inconsistent methodological approaches used in the data 
         generating process. 
   The harmonisation of processes used in the reporting units and in the
   NSIs is also of decisive importance for consistency and thus for
   statistical quality in a system wide context.
                                                                         4
         Norbert Rainer, Josef Richter                                Vienna, June 2014                                                                                   4
             Q2014 european conference on quality in official statistics
        Special session: Consistency of Concepts and Applied Methods in Business Statistics
   Special characteristics of business statistics
   Institutional background of the fundus of information available to 
   respondents. 
   High degree of division of labour in the various steps of the data
   generating process:
         Step 1   Basic observations at the unit level
         Step 2   Classification and aggregation of basic observations at the 
                       unit level; design of the information system
         Step 3   Statistical reporting to the NSI
         Step 4   Statistical editing in the NSI
         Step 5   Classification and aggregation of micro data by the NSI
                                                                       5
         Norbert Rainer, Josef Richter                                Vienna, June 2014                                                                                   5
          Q2014 european conference on quality in official statistics
      Special session: Consistency of Concepts and Applied Methods in Business Statistics
   Important sources of inconsistencies in the data 
   generating process
   Inconsistencies on the level of single respondents
       Concepts -   Use of different definitions in different statistical domains
       Processes - Different codification of identical products by different people
   Inconsistencies on the micro level of micro data between units
       Concepts -   Use of different statistical units in different statistical domains
       Processes - Different methods of decomposing’ basic information
   Inconsistencies on the level of Statistical Offices - edited micro data sets 
             Processes - Use of different methods of editing and imputation across
                                 domains
             Processes - Lack of system wide checks for consistency 
    
                                                       6
         Norbert Rainer, Josef Richter                                Vienna, June 2014                                                                                   6