Government needs to eat its own data

In Whitehall Monitor 2015: the Coalition in 163 charts, we talk about data in three ways:

• data as ‘data’ - it is complete, consistent and accurate
• data as ‘information’ - it says something meaningful, and helps us to understand what is happening
• data as ‘evidence’ - it can be used to inform decisions.

Jeni Tennison used the analogy of driving a car without looking at the fuel gauge: you might find yourself running out of petrol in the middle of nowhere. By the same token, government needs to use its own data to understand how it is performing, and where problems may arise. This is critical in the context of the tough Spending Review settlements due to be published this week: decision-makers in government will need good data to understand the resources they have, and the consequences of making reductions in different ways. This is why the panel’s anecdotes about poor quality, missing and inconsistent data are important: not to bash the poor soul(s) who put the data together, but to show that it is not being used. Everyone has to accept there will be mistakes in their data (one of the five stages of ‘data grief’) but getting more people using it increases the chance of those errors being spotted and corrected. The term used by Matt Hancock, Minister for the Cabinet Office, at the Open Data Institute Summit last week, was ‘dogfooding’: like the apocryphal dog food magnate who insisted on eating his own product in front of his shareholders,  government should publish data of such a high quality that they are prepared (and able) to use it themselves. Jeni Tennison talked about ‘data infrastructure’: the governance models, standards and lines of responsibility that need to sit around data sets if they are going to be useful. Developing good data infrastructure involves making often difficult, central decisions about what data to collect, how it should be published, and who is responsible for maintaining it. At the moment, different government organisations often don’t trust other organisations’ data, so they collect their own. Good data infrastructure – for example, a single canonical register of all government organisations – allows different organisations to work together more effectively, and helps everyone to understand how different parts of government compare and interact. Data infrastructure includes the people in the organisation who know how to collate and interpret data properly. Megan Lucero said her expert team had enhanced The Times’ use of data. They spotted the problems with polling data months before the election, and made the difficult decision to stop using it. Instead, they used an algorithm that drew from other data sources, and were able to learn as the results came in. Improving data infrastructure will make data more comprehensible, but it will not ensure that the data is useful. Several audience members asked about data on outcomes: how do we actually measure the performance of government in making an impact on the world? Earlier this year, the Whitehall Monitor team attempted to chart the performance of government according to the ‘impact indicators’ named by departments as ways to assess their departments. We found this regime to be inadequate, with no baselines or benchmarks allowing us to assess performance. Jeni Tennison echoed the Institute for Government’s hope that Whitehall's new Single Departmental Plans will contain a clear metric for each objective they contain.

Data journalism is nothing new, says @Megan_Lucero. Here's @thetimes GE graphs from 1895 and 2015 #WM2015pic.twitter.com/UKV4OdctFM

— Jo Suherman-Bailey (@JSuhermanBailey) November 19, 2015

Megan Lucero raised the important point that data visualisation and analysis is not new. The Victorians were big fans of trying to collect and understand data in order to make decisions and understand the world. It is even more important, then, that we do not rest on our laurels and allow government data to carry on being used and published in a varied and intermittent fashion. Ian Makgill of Spend Network asked how we can encourage government to use and publish better data. The panel’s answer? To show what we can do with it. The Whitehall Monitor team will continue to do just that.