Move fast and don’t break things: the challenges of strategy for statistics

Ed Humpherson, Director General for Regulation, writes about the challenges faced by statisticians formulating strategy in a constantly changing landscape.

At the end of last year, I attended a conference where heads of statistical offices from various Commonwealth countries put their heads together. One session involved them defining the leadership challenges statisticians face.

Their list said that statisticians in leadership roles need to:

  • demonstrate high levels of integrity
  • be deeply knowledgeable about statistics
  • have the ability to set vision and define strategy
  • be skilled in communication
  • have strong people leadership skills

We also heard they need resilience; need to be inclusive; face increasing demands and loss of monopoly; and need to build partnerships across society. These challenges are significant, and not just for heads of statistical agencies, but for statistical professionals and leaders across Governments.

But to some extent the list is quite generic. It’s difficult to think of a role that doesn’t require communication and people leadership skills; integrity is – or should be – a valued commodity in every role. What is it that is unique to statistical leadership in official statistics?

The challenges of strategy for official statistics

To me the most salient of these challenges revolves around strategy. Statisticians face an environment that is changing and full of a series of trade-offs:

  • being expected to provide definitive views versus recognising the uncertainty in the statistics produced: statisticians are often asked for The Number, the single point estimate that resolves an issue – or can be dropped into a briefing or speech as it if was just padding, mere upholstery. But the statistical leader knows that the single number is elusive; whether it means what people think it means depends crucially on what question you are trying to answer. Conveying this while still providing relevant, useful insight, is a huge challenge.
  • using new forms of data and new tools versus providing consistent time series. The expectations that statisticians get with the worlds of Big Data and data science are growing all the time. Yet because new data can create new insight, they can also represent a decisive break with the past. So what should the statistician do? Preserve a time series so the present can be compared with the past? Or update methods and data sources to take advantage of all that the digital revolution has to offer? Our answer at the Office for Statistics Regulation starts from the principle that statistics should always be the best available estimate. But it’s not straightforward for statisticians to deliver on this principle where stakeholders value and need consistency of data over time. The best balance is likely to differ in different contexts and for different users of statistics.
  • serving experienced users versus wider populations of ordinary citizens. “Users” is a broad and complex concept. What works for those who delve deeply into the data, use it regularly to answer specific technical questions, may not work for people who only rarely engage with the statistics. This is the challenge of the lived experience, of the “that’s your GDP, not mine”, and it’s a huge issue for those who produce statistics.
  • serving Government decision makers like central banks and finance ministries versus informing wider public debate. Statistics must serve a much wider constituency than this – and serving both the elite decision making group and the wider audience can be demanding. A strategy like Better Statistics, Better Decisions can looks like it’s focused on senior policy makers who make the Big Decisions, even though John Pullinger (the National Statistician) always emphasises that the idea of a “decision maker” can go much wider.

The common thread of these trade-offs is managing a balance of technical and democratic roles. Statisticians are technocratic – perhaps the epitome of the technocrat – but they also perform an essential social role: the public want to see themselves, their lives, in the statistics, and not feel that statistics are remote, technical, abstract, dry. And this balance is harder now than ever – because the availability of data and the complexity of our societies demands more sophisticated tools of analysis (more technocracy) at the same time as there is a growing sense of mistrust of evidence, data, elites, and a growing demand among people for recognition of their identity (more social engagement and democracy). All of this requires openness: opening up while preserving core capabilities.

And supporting this opening up is such an important role for the Office for Statistics Regulation. We do this when we assess statistics – always emphasising the central role of opening up to a wide range of users. We do it when we do systemic reviews, focused on the value of statistics in areas like housing, health and migration. And we do it through our Code of Practice for Statistics, which is all about opening up the processes and outputs of statistics.

Move fast and break things?

Statisticians need to embrace disruption, and be urgent – to be ambitious and move quickly. But this disruptive task is not as simple as the motto of a commercial tech company: to move fast and break things (apparently Facebook’s original motto).

The strategic challenge is much harder, and therefore more interesting: the trustworthiness of official statistics is incredibly precious and statistical producers hold it in trust for the societies they serve. They mustn’t casually destroy it. This is what makes it an energising challenge – to change and develop statistics using new data sources and new techniques while enhancing connection with the public that statistics exist to serve.

And it’s our job is to help producers of statistics meet their strategic challenge: to Move Fast But Not Break Things.