|
The sample survey has been central to the creation of government statistics for decades. The now well-known problems of declining response rates and rising costs have led many to argue that the era of the survey is over. The increasing availability of alternative data sources (administrative records, "Big Data") reinforces this claim. A recent report of the Committee on National Statistics seeks to foster alternatives to the "survey paradigm." It urges federal statistical agencies to research and invest in combinations of data from non-survey sources to create the statistics that have been gleaned from sample surveys heretofore.
Is it time to abandon the sample survey? Are nonresponse and rising costs intractable problems, as some allege, or are they issues susceptible to new ways of thinking? On the other hand, can administrative records and "Big Data" be shaped into products that satisfy the requirements of government statistics and the needs of data users who are accustomed to what surveys provide? Participants in this roundtable will discuss such questions and debate the future of the sample survey for government statistics.
|
