Many statistical organizations are considering tranformation of their production base from sample surveys to the integration of surveys with other data sources, e.g., administrative and commercial records, or some forms of "big data." That transformation can lead to stakeholder concerns about "break in series" phenomena arising from changes in, or loss of, one or more data sources. This roundtable discussion will cover some practical aspects of these problems, including:
(1) Definitions of "breaks in series" and related problems, including both readily observable breaks and more subtle phenomena like changes in dispersion or seasonality patterns
(2) Methods for timely identification of these problems
(3) Methods for mitigation of the impact that these breaks have on data users
(4) Incorporating (1)-(3) in decisions about whether to use a prospective data source for production of a statistical series
|