Chapter 8 Final Words

There are two important lessons which must be drawn from this course:

  • Sample surveys are a key source of data in many different domains, including economy and finance. Contrary to macro sources, surveys provide indicators on distributional aspects such as income or wealth inequality measures.
  • Sample surveys is a serious business, which must be carried out properly following scientific guidelines with the aim to produce quality figures and communicate about them.

Even though they are an essential instrument to obtain relevant statistical information, sample surveys now face important challenges:

  • The constant decline of survey participation, which poses a serious threat to sample representativity and overall to the quality of survey results. Higher non-response is caused by many factors such as the increasing number of survey studies conducted not only by National Statistical Institutes, but also by private organisations, universities, research centres etc., and their lack of coordination. There are also growing fears among populations regarding data privacy and how personal data are treated. People are also more mobile and therefore more difficult to catch. This aspect is particularly true in a country such as Luxembourg, where half of the resident population is composed of foreigners.

  • The need to provide statistical indicators more and more quickly (timeliness), which implies the development of data production chains and the growing need of IT skills among the staff.

  • The integration of new data sources. For instance, administrative databases may provide a great deal of accurate information on household income, which can be used as an alternative to income data collected from sample surveys. “Big data” sources (Facebook, Twitter etc.) also raises growing interest among data users although there are important methodological questions with regard to their quality, their representativity and their access.

  • The use of new digital tools for data collection, for instance internet-based surveys, mobile apps, smart meters, smart cards etc. Digital tools offer interesting solutions to facilitate data collection, reduce survey burden and help reduce errors, but they are not easily accessible to every individual in a population (see e.g. “digital gap” between the young people and the elderly)

Notwithstanding all this, sample surveys are not yet doomed by the rise of digitalisation. There are plenty of topics (e.g. subjective questions) which are not addressed by “Big data” sources and for which survey data are still relevant. However, the future of Statistics should probably go towards more combination of data sources and more mixing of data collection modes (telephone, web, face-to-face etc.)