Government needs to update the way it uses data in policymaking - Photo credit: Flickr, Janet McKnight

Speaking at an event on big data for local government hosted by the Oxford Internet Institute this week, the institute’s director Helen Margetts said that there was much more that could be done with the data generated by government.

A long-running characteristic of government in the UK, she said, was that the data government generates is “completely un-extractable” in any form that can be used for service design.

“The inability to extract data for policymaking [causes government to] miss opportunities that are there,” she said.

Related content

Local leaders told to 'grasp the nettle' in big data analytics, says Socitm
Bringing government data to life

Margetts argued that there was an increasing amount of pressure on government to use data for policymaking, but that it needed to think about data in ways that it hadn’t before.

For instance, she said that government’s method of collecting official statistics “properly, year on year, in the same format” means it is less open to other forms of data.

“You don’t find it in the wild,” she said. “I think that’s a cultural barrier in government for data science of big data.”

There is also a lot of data “bubbling up in society” that government could be making better use of, she said. “That data can be used to make the services we use better.”

Further pressures include people’s expectation that they will be able to interact digitally with services and the fact that many things that weren’t digital.

“We need to move away from the idea that [data science] is a frivolous thing, that it's a nice-to-have extra for geeks, and towards it being something that’s unavoidable; something that has to happen,” she said.

Margetts also said that, quite often, organisations “know very little” about their data or how much data science expertise there is within the organisation, and urged them to map internal data sources and expertise.

She also urged government bodies to link old and new data sources. “There are all sorts of data that’s never been available before,” she said, “but that doesn’t mean we don’t want to link to older data.”

Traditional opinion polls and census-style surveys will have a lot of demographic information attached to it that transactional data doesn’t.

“If you’re just looking at Google analytics, you might know where [a user has] come from or where they go, but that’s it,” she said. “Linking old and new data will be a method for the future.”