The lunch starts at 12:00, and the talk will start around 12:20.
NB: Wednesday BigInsight Lunches are open to staff and students from any of the BigInsight partners, including UiO, but not to others.
Speaker: Li-Chun Zhang, Researcher at Statistics Norway and Professor II in statistics at UiO.
Location: Spiseriet, Norwegian Computing Center
Title: How to use big data for Official Statitics?
Abstract: Traditional survey sampling has many attractive features that are particularly important to Official Statistics, such as transparency, universality and flexibility. Over the past decades reusing data originated from administrative sources has become common in Scandinavia; the momentum is fast growing internationally. The most recent digital revolution has greatly improved the availability of so-called big-data outside the administrative sources, which can as well be repurposed for Official Statistics. Against this background, I will outline some key overall issues one needs to resolve, so that big data can both enhance and enrich the Official Statistical system. As examples of specific methodological issues, I will discuss the identification of precise validity conditions for descriptive inference from non-probability big-data sample and the difficulties of empirically validating these validity conditions, the adoption of network representation of the relevant data and the associated opportunities and problems. Transaction and cellphone data will be used to for illustration. Despite the best effort, bias may be unavoidable and may dominate the variance in many situations, which creates obstacles for statistical inference. An approach to uncertainty assessment in the presence of such big-data bias will be outlined and illustrated.