Thoughts from the mind of Ben Welby

Tag: data

The Path to Becoming a Data-Driven Public Sector

This paper was a team effort under my leadership with Arturo contributing the chapter on data governance, Lucia working on data for trust and my developing the introduction, conclusion and the material around data for public value.

The bones of the framework are being used in our Digital Government Reviews and it is our hope that you could take the 12 elements of the framework and apply it into any context. Indeed, there are two appendices to the report that, thanks to our colleagues Gavin and Daniel, apply the framework to the context of 1) integrity actors and 2) human resources.

If you’re interested in consuming this report in a different format then I’ve posted the series of seminars I hosted with the Azerbaijani government in May 2024 which unpacks the Framework as a set of presentations.

Available as HTML or a PDF

What’s the TL;DR?

This report builds on our earlier working paper by doing two things:

  • It introduces country-level practices and insights provided by several OECD member countries that contribute to the E-Leaders Thematic Group on Data
  • It uses those insights to develop a framework setting out the steps that countries need to take in order to build out an effective approach to the data-driven public sector

With ‘Data-driven public sector’ being one of the six elements which we argue form the basis for digital government maturity it is incredibly important for governments to address all the elements that go into achieving maturity in this regard.

To that end the framework consists of three pillars:

Pillar 1: Governance: we cast the vision for ‘governance’ wider than legislation, regulation and responsibility for data which is what people tend to understand this means. We argue that effective governance involves strategy (leadership), tactics (implementation and rules) and delivery (infrastructure and architecture).

Pillar 2: Public value: the point of putting data to work is to meet user needs and deliver societal value. We draw on country practices to show how important data is to looking ahead to future (anticipating and planning), responding to immediate needs (delivery), and then understanding what can be learnt from the past (evaluation and monitoring).

Pillar 3: Public trust: it is far easier to lose trust than it is to build it. That means governments need to be thinking about all the ways in which the use of data could undermine public trust. We explore dimensions of ethics, privacy, consent, transparency and digital security.

In our work we find that governments may hive off different elements of this under different organisations and while they may have good plans and practices in place, often there is not a holistic and strategic overarching sense of how these elements interact. The starting point has to be strong strategic leadership, but that leadership must be mindful that there’s almost nothing that can be achieved with data in the public sector without making public trust the guiding priority.

A circular diagram divided into three equal segments labeled 'Governance,' 'Public value,' and 'Trust.' Around the 'Governance' segment are five boxes: 'Leadership and vision,' 'Coherent implementation,' 'Rules and guidelines,' 'Data infrastructure,' and 'Data architecture.' Around the 'Public value' segment are three boxes: 'Anticipation and planning,' 'Delivery,' and 'Evaluation and monitoring.' Around the 'Trust' segment are four boxes at the bottom: 'Ethics,' 'Privacy and consent,' 'Transparency,' and 'Security.'

The blurb

Twenty-first century governments must keep pace with the expectations of their citizens and deliver on the promise of the digital age. Data-driven approaches are particularly effective for meeting those expectations and rethinking the way governments and citizens interact. This report highlights the important role data can play in creating conditions that improve public services, increase the effectiveness of public spending and inform ethical and privacy considerations. It presents a data-driven public sector framework that can help countries or organisations assess the elements needed for using data to make better-informed decisions across public sectors.

Available as HTML or a PDF

York’s local election

In the aftermath of York’s election I was interested in the sort of things that might get talked about at a general election in terms of a picture changing from election to election. The simple picture was a crushing defeat for the Liberal Democrats as Labour swept to power but in the spirit of Whitehall Watch is there a story in the votes rather than the seats?

First up, the makeup of the chamber before (taking into account the by-election results since 2007):
City of York Council 2007

After:
City of York Council 2011

And you can see from a quick glance how the vote changed from 2007 to 2011

Labour recorded 18,000 votes more on Thursday than they did in 2007 and increased their share of the vote from 27% to 37%. Much has been said about Thursday representing an incredibly bad night for the Liberal Democrats but in York the result was not down to the complete evaporation of Liberal Democrat support. The ruling Liberal Democrats (only) lost 5,000 votes across the city, a similar figure to that dropped by the Conservatives. The other party to shed votes was the BNP whose support shrank by 70%.

Party Votes 2007 Votes 2011 Change
BNP 3,582 1,076 -2,506
Conservatives 37,172 32,788 -4,384
Green 14,337 19,196 4,859
Labour 36,746 54,874 18,128
Liberal Democrats 43,764 38,818 -4,946
Others 829 1112 283

Swing is a favourite statistic to work out and in order to calculate it you take the increase in votes from one party, add it to the fall of the other and divide it by two.

This means a swing to Labour from the Liberal Democrats of 8% with a similar figure for the swing to Labour from the Conservatives of 7.7%.

But there are some more nuances to what actually happened in the city. Although the Conservative party vote fell by 5,000 this is more connected to a reduction in candidates from 47 to 33. In 3 wards, which had contributed 3,724 votes in 2007 there was no Conservative candidate at all. Despite their share of the city’s vote falling to 22% their average vote per candidate increased to 994, more than any other party except Labour.

In contrast, the Green party’s significant improvement overall comes from their fielding an additional 15 candidates. In Clifton, for example, they tripled their candidates and secured an additional 10% of the vote (although their leading candidate only increased her votes by 33). But, apart from Skelton, Rawcliffe and Clifton Without (the neighbouring ward) where they picked up almost 12% more of the vote their performance across the city was fairly static with the average number of votes each candidate received falling by 74.

The Liberal Democrats don’t seem to have simply lost seats due to dissatisfaction with the national political picture. In Strensall, Haxby and Wheldrake they lost seats to their coalition partners (Christian Vassie’s loss of Wheldrake compounding a miserable 12 months in politics after last year’s general election when he was unable to dislodge Hugh Bayley).

So the convincing nature of Labour’s victory seems to be much more down to getting people to vote rather than seeing a massive drop in support for either the way the Liberal Democrats ran York or in the Conservative-Liberal Democrat coalition in Westminster. In 2007 the turnout was 41.8% (approx. 136,000 voters) but this year the extra 10,000 voters in a turnout of 44.7% seems to have made all the difference rather than disgruntled voters switching from one party to another. Edit: of course, I’d failed to think about the aggregating effect of wards where individuals got more than a single vote. The difference between 2007 and 2011 was actually closer to 4,500 voters.

Mind you, given that the majority of York wasn’t voting last Thursday I wonder whether any of these thoughts are in any way relevant.

If you want to pick over the data and point out any of the flaws in my data literacy there’s a spreadsheet on Google Docs.