A. Summary of good practices
21.2. Compilers are encouraged to consider implementing such electronic data collection methods as electronic questionnaires, computer-assisted personal interviewing, Internet data collection and electronic file transmission, as they offer the potential for improving the accuracy and timeliness of statistics, reducing survey cost and reporting burden and diminishing the compilation burden. Given the complexities of collecting data on the international supply of services and its various dimensions, adopting an electronic data system may assist in collecting detailed breakdowns by type of service, partner country or mode of supply, as well as for different types of variables.
21.3. It is good practice for electronic data collection methods, especially electronic surveys, to include the use of built-in automatic edits and other devices, such as automatic data fills and calculations, prompts for missing fields and automatic skipping of not-applicable questions, that allow respondents to avoid errors and to fill in the questionnaire faster and more easily.
21.4. Compilers should also be aware that the use of electronic questionnaires may introduce a certain bias in survey results (at least in the early stages of implementation), because some potential respondents may be unable to participate owing to a lack of access to, or a lack of familiarity with, the appropriate technology.
21.5. Compilers are advised to design their processing systems in an efficient way so that data and metadata can be conveniently retrieved from the relevant databases, be used in the generation of the intermediate and final data sets and be updated and synchronized. In that context, compilers are encouraged to design and use a warehousing system for data and metadata to integrate the dissemination of information with the collection and processing components of the statistical production process.
21.6. Compilers are further advised to choose the form of technology used for dissemination on the basis of the nature and quantity of data to be published. In making such decisions, statistical organizations should also consider the needs of their users, e.g., publish a standard data set that meets the needs of most users via the most readily accessible technology, and provide more sophisticated data sets via different dissemination methods, such as online interactive databases (free or not). Printed publications can also be produced more easily with an effective use of information technology.