Robert Groves, Provost of Georgetown University, Nancy Potok, the Chief Statistician of the United States with the Office of Management and Budget, and Tjark Tjin-A-Tsoi, the Director General of Statistics Netherlands (from left), discuss statistics and big data.
What are the possibilities that come with big data, and how can governments best develop policies based on the wealth of data available?
That question was at the center of a discussion between Tjark Tjin-A-Tsoi, the Director General of Statistics Netherlands (CBS), and Nancy Potok, the Chief Statistician of the United States with the Office of Management and Budget, at the McCourt School of Public Policy of Georgetown University on February 7. Robert Groves, Provost of Georgetown University, led the discussion.
Also present were representatives of statistical agencies, Georgetown staff, and members of the Commission on Evidence-Based Policy Making, a bipartisan organization committed to strengthening the government’s evidence-building efforts.
While both Statistics Netherlands and the US statistical bureaus provide important data and statistics for the government, Potok and Tjin-A-Tsoi explained they are not and do not wish to be directly involved in policymaking. Instead, they acquire and publish data on which legislators can base their policies.
Despite their similar objectives, there are a few important differences between the way Statistics Netherlands and statistical bureaus in the US operate. Firstly, the US has 13 decentralized statistical agencies, each tasked with researching their own field.
The Netherlands, on the other hand, only has one centralized bureau, which is Statistics Netherlands. This allows the Dutch to quickly combine data from different sources without the need for outside approval. For the US statistical bureaus this process is more cumbersome and tedious.
The public perception of the respective bureaus also differs. In the Netherlands, the public’s most important concern is ensuring that Statistics Netherlands remains politically impartial. In the United States the main issue is the public’s skepticism toward letting the government gather private data about individuals.
Lastly, Statistics Netherlands is ahead of many of the US statistical bureaus when it comes to making its research available online. Statistics Netherlands has had its own newsroom for many years, and publishes new data on their social media accounts. Dutch media outlets often pick up these publications.
Larger US statistical bureaus, such as the Census Bureau and the Bureau of Labor statistics, have the resources to conduct robust communications and data dissemination efforts to make their data and research available. Many of the smaller bureaus, however, do not have the resources to to go beyond smaller efforts.
A competitive edge
To use the publicly available data for evidence-based policymaking, it is important to integrate traditional sources of data with new sources, said Tjin-A-Tsoi and Potok.
Data has traditionally been acquired through high-quality, structured censuses and surveys. In contrast, new sources of data, such as information collected through automated sensors or scanners, are not as structured. However, this kind of data can be collected faster, and in greater variety and amounts.
The integration of two data sources would give statistical agencies such as Statistics Netherlands a competitive edge over private agencies because, as a centralized and public agency, it still has access to data that companies such as Google do not have, Tjin-A-Tsoi said, and we should use that to our advantage.
Potok agreed. “There is no sentiment for eliminating traditional data,” she said. “The demand for it will always remain. Our job is to integrate it with other sources of data.”
This would allow official data releases to be more timely, be available at lower levels of geography, and potentially save money over the cost of conducting surveys at a time when public cooperation is decreasing.
Tjin-A-Tsoi acknowledged that policies based on big data “won’t solve all the world’s problems at once. Trying to do so will stifle all innovation. Instead of trying to confront a giant problem upfront,” he said. “Organizations should tackle smaller cases first and learn from that.”
“As with anything new, errors will be made,” Potok added. “But we have to find our way in it. The fact that there is a risk to something doesn’t mean you shouldn’t do it.”
One of the ways to mitigate the risks of creating new data products by combining data from multiple sources is to have a plan B for data acquisition and never rely on only one data source, she said.
Both Tjin-A-Tsoi and Potok agreed that there is high value for the public to be gained from the new wealth of data that has become available to statistical agencies in recent years. For the data to have value, however, “telling the stories behind the data is important,” Tjin-A-Tsoi said. “People don’t remember facts. They remember stories.”
Tjin-A-Tsoi said, “publishing data without any guidance or backstory will lead to misinterpretation and misuse of the data. Statistical research needs to answer to society’s demands.”
He described Statistics Netherlands as a “giant detector” that uses “targeted experimentation” to prove or disprove hypotheses. Potok agreed, but added that statistical agencies need to be wary not to “make the data politically biased.”
A close partnership
In response to a question from an audience member, Tjin-A-Tsoi explained that the reason for their short visit to the US was to start building a close partnership between the Netherlands and the US.
Despite being one of the most advanced national statistical bureaus worldwide, “Statistics Netherlands can’t do it alone,” Tjin-A-Tsoi said.
To this end, a memorandum of understanding was signed between Tjin-A-Tsoi and the Provost of Georgetown University, Robert Groves. They expressed their intent to share their expertise via mutual visits to each other’s headquarters.
During their two-day visit to the US, Tjin-A-Tsoi and Marcel van der Steen, Manager Innovation and Business Development at Statistics Netherlands, also met with other organizations, such as Westat, IBM, the Pew Research Center, and the Bipartisan Policy Center, to discuss the future of big data and evidence-based policymaking.