Eurostat Data collection
Presented by Johan Erikson Statistics Sweden
Data collection - outline Modes of data collection Choosing the mode(s) Mixing modes Choosing the contact strategy Setting up the collection – web questionnaires Data collection modules in the handbook
Modes of data collection Interviewer-administered modes –Telephone interviews (CATI) –Face-to-face interviews (CAPI) Self-administered modes –Paper questionnaire –Electronic questionnaire (CAWI, downloadable forms, and different types of file transfer)
Modes of data collection (2) Modes without respondent –Electronic data interchange (EDI, XBRL) –Observation –Administrative data (secondary data)
Choosing the mode Factors to consider: –Response process –Response burden –Desired data quality (response rate, measurement error) –Available resources Budget Staff IT-resources and respondent´s technical conditions –Respondent expectations –Topic of the survey –Timetable
Response process Assigning priorities Authorisation Data in records Gatekeepers More than one respondent Based on this, self-administered modes are most common in business surveys
Response burden Questionnaire design vs data availability Timing Finding the right person(s) and dividing tasks Questionnaire design is vital!
Data quality Response rates are generally higher in interviewer-administered modes However, interviews might not be an alternative And business surveys are often mandatory Helping the respondent to higher quality –Pre-printing –Built-in functionality in EQ
Mixing modes Using more than one mode Costs of mixing modes: –Several questionnaires, set-up costs –Risk of mode effects –Staffing Advantages of mixing modes –Coverage –Response rates and data quality –Costs in the collection stage
Mixing modes Two main types of mixed mode Parallel approach –Lets the respondent choose Sequential approach –Primary mode offered first –Typically, cheapest mode first A third approach: Differentiating modes: –The most expensive modes not used for all businesses
Most common data collection strategy: Use already collected data when possible Promote the main mode (web) all the way Design for mixed modes Use modes in sequence, starting with the cheapest Optimise the questionnaire design for each mode Have special treatment of the largest businesses
Contact strategy When to send –Enough time to finish the survey –Data availability –Reasonable time to respond –No. of reminders –Other surveys What to send –Based on the mixed mode strategy –Paper form or not?
Contact strategy (2) Method of contact –Mail (letter) – –Telephone Who to contact –Reaching the best respondent –Difference between small and large enterprises –Using previous contacts –Using a generic role
Contact strategy (3) New businesses in the survey –Specific treatment or not? –Sample rotation, ongoing surveys –Finding the right person –Showing the enterprise the importance of replying
Contact strategy (4) Specific treatment: –Single point of contact –Building relations –Profiling statistical units –Finding contact persons –Support for the enterprise –Support for survey staff –Enterprise-specific arrangements –Coherence analysis
Contact strategy (5) Feedback to respondents –General feedback –Enterprise-specific feedback Responsive design issues –Making design choices based on results so far: Switching modes Switching contact methods Different questionnaires Penalties/fines
Set-up collection: Web questionnaires Introduction letter –Describing the survey –Letting the businesses prepare Login procedures –Security issues –Must be easy
Web questionnaires (2) Editing during data collection –Hard errors / soft errors –Client-side / Server-side Platforms and browsers Response time Several response sessions Web-site / Business portal
Web questionnaires (3) More advanced electronic alternatives: –Touch-tone data entry –File transfer –File import (XBRL) –Electronic data interchange
Data collection modules Main theme module - overview Design of data collection 1 – modes Design of data collection 2 – contact strategy Techniques and tools Mixed mode CATI allocation Collection and use of secondary data