Big Data and Social Indicators: Actual Trends and New Perspectives

Big Data are a top subject in international research articles and a vast debate is taking place on their actual capability of being used to complement or even substitute official statistics surveys and social indicators in particular. In this paper we analyse the metadata of the Scopus database of academic articles on Big Data and we show that most of the existing and intensively growing literature is focused on software and computational issues whilst articles that are specifically focused on statistical issues and on the procedures to build social indicators from Big Data are a much smaller share of this vast production. Nevertheless the works that focus on these topics show promising results because in developed countries Big Data seem to be a good information base to create reliable proxies of social indicators, whereas in developing countries their use (for instance using satellite images) may be a viable alternative to traditional surveys. However, Big Data based social indicators deeply suffer of a number of open issues that affect their actual use: they do not correspond to any sampling scheme and they are often representative of particular segments of the population; they generally are private process-produced data whose access by national statistical offices is rarely possible although the intrinsic value of the information contained in Big Data has a social importance that should be shared with the whole community; Big Data lack the socio-economic background on which social indicators have been founded and their help to policy makers in their decision process is a fully open point. Therefore Big Data may be a big opportunity for the definition of traditional or new social indicators but their statistical reliability should be further investigated and their availability and use should be internationally coordinated