This paper provides a survey of economic research using alternative data in Japan. The use of alternative data in economic research has increased signi.cantly since the outbreak of COVID-19 in early 2020s. A large number of studies have been conducted in a wide range of areas, including i) nowcasting, ii) analysis of changes in economic behavior and in economic structure, and iii) measurement and evaluation of policy e.ects. These stud-ies have proved that alternative data are particularly useful when economies exhibit high volatilities. More recently, the .eld of research is expanding toward topics not necessarily related to COVID-19. These developments of research, however, have also revealed caveats and challenges from the point of statistical reliability, when using alternative data. These issues include insu.cient knowledge about basic characteristics of data and its relationship with o.cial statistics, and degree of adequacy of sampling method and sample design.

Since 1955, the Asahi Shimbun has been conducting national pre-election polls, evolving from face-to-face interviews to telephone surveys sampled from voter lists and eventually to the current hybrid method combining Random Digit Dialing (RDD) for landlines and cell phones with internet online surveys. This paper traces the history of these methodologies and explains the methodology of RDD and Internet survey showing the challenges for them. It also highlights the signifcance of exit polls, which have been experimentally conducted since the 1990s and have become indispensable for election reporting on television and vote analysis in newspapers.

In the field of marketing research, in addition to the traditional market research, digital data analytics are required, and the insight business that utilizes various data is expanding. Big data used in digital data analytics can be collected in real time and often serves as an alternative to factual surveys using questionnaires. Questionnaire-based consciousness surveys are an effective means of understanding the level of interest in a theme. However, even if the number of consciousness surveys, which are often conducted annually, is increased to enhance their real-time performance, it is difficult to conduct them daily because of the large burden of responses and cost, even monthly surveys are not feasible. Therefore, we focus on the easily obtainable Wikipedia data and investigate their usefulness as digital data in reflecting individual interests. In this paper, we organize the characteristics of Wikipedia data from the viewpoints of data type and location and propose a method to illustrate changes in the number of accesses to Wikipedia pages. Using this method, we grasp the change in public consciousness from the number of accesses to article pages that consumers were interested in during the COVID-19 pandemic. We demonstrate the role of Wikipedia data as consciousness data and not as behavioral data.