Flashy innovation will not fuel the data revolution
- Put national statistics offices at heart of each country’s data shake-up
- They need support to integrate partnerships and technological innovations
- Improving statistical systems takes time and money
Calls for a data revolution to inform the post-2015 sustainable development agenda have been met with commitments, made at the Addis Ababa Financing for Development summit last month, to invest in national statistical systems, make greater use of unofficial data including big data, and adopt technological innovations to support data collection, analysis and dissemination.
These could all fill existing data gaps, improve public services and broaden partnerships. But as commitments turn into on-the-ground initiatives, the data revolution must become rooted in national priorities and realities.
And rather than the technological innovations that claim the data revolution spotlight, in many countries this will require something less flashy: considerable investments in long-term statistical infrastructure and capacity development.
The Post-2015 Data Test project, which I colead, was set up to examine the adequacy of data for measuring progress in several development sectors. And, so far, testing in a handful of countries shows that many continue to face bottlenecks in statistical infrastructure and capacity development.
In Bangladesh, for example, data storage is not fully digitised, and weak information and communications technology (ICT) infrastructure hinders data collection, analysis and dissemination. In Sierra Leone, limited internet access and use of smartphones or tablets are the main barriers to people accessing official statistics disseminated online. And, in Tanzania, only 15 of the 150 staff at the National Bureau of Statistics are proficient in the use of statistical computer programmes, while other government departments are also short-staffed.
Other issues are common to many developing countries. For example, data collection across government departments is often uncoordinated with national statistics offices (NSOs). And it often fails to meet quality standards or use consistent methodologies and definitions.
So how to strengthen statistical systems? Bill Anderson, a data expert at the NGO Development Initiatives, says this can’t be done through quick-win, ‘plug and play’ interventions. His work in Uganda on joining up disaggregated data sets to create usable, highly localised information highlights this reality.
Results can only be achieved by taking a long-term view to improving statistical capacity. For example, in Senegal, the challenge of retaining senior statisticians led to the government-backed National School of Statistics and Economic Analysis. The school is linked to the NSO, which offers pre-service and in-service training. The move has been successful: since the school was established, it has contributed to the NSO’s recruitment and retention of trained statisticians, and overall staff numbers rose from 106 in 2000 to 234 in 2010.
In short, efficient, effective systems that produce sound official statistics do not appear overnight. They require physical, human and technical resources in NSOs and government departments.
Long-term support will also enable NSOs to harness the potential of innovative technologies and unofficial data, which can address immediate gaps. This is because innovations will need to be grounded in country-level realities. In practice, this will mean understanding NSOs’ resource constraints before settling on appropriate innovations.
“Efficient, effective systems that produce sound official statistics do not appear overnight. They require physical, human and technical resources in NSOs and government departments.”
Choosing technologies carefully can lead to huge gains in data availability and quality. Senegal used personal digital assistants in its 2013 census following a successful trial. This meant that preliminary results were available within three months — a vast improvement compared with a five-year lag in the previous census. It also enabled the collection of data with greater levels of break down by location, age and sex.
In this instance, the new technology succeeded because it had been tested, had sufficient funding and matched staff capacities.
Effective support for NSOs
There are new ideas for how to support NSOs effectively.
Many observers have called for better coordination with unofficial data producers to fill gaps in official statistics. Such partnerships will require NSOs to ensure data quality and coordinate collection — but these capacities are already weak.
And although use of unofficial data can offer quick wins, it should not be used at the expense of strengthening institutional capacities. NSOs can make better and greater use of the data they, and other departments, already collect.
International data partnership PARIS21 and other commentators have also suggested using data compacts, where countries agree to a set of basic principles and minimum standards in exchange for external financing. Such a system could improve data quality and boost financing, which is crucial: Post-2015 Data Test studies point to a lack of timely funds as a key obstacle to strengthening statistical systems. Even when national strategies are in place, implementation often lags from delays in financing.
In the past, donors chose to fund statistical activities that reflected their priorities rather than national plans. But future investments from governments and the international community must match national priorities.
Commitment to the proposed data compacts may facilitate this, if activities and partnerships are planned according to realities on the ground. This would help ensure that the data revolution is dominated by local rather than global needs.
As stewards of official data, NSOs should be at the heart of each country’s data revolution. But for their efforts to be sustainable and relevant, they need support to identify when, where and how unofficial data can fill data gaps, and to integrate technological innovations and partnerships into a statistical system that aligns with national priorities.
Shannon Kindornay is an adjunct research professor at Carleton University, Canada, and coleads the Post-2015 Data Test. She can be contacted at @skindornay and @post2015data
This is part of a set of pieces on data funded by the Hewlett Foundation.
What else do you want to know? Tell us what you would like us to investigate about the data revolution.