About

What is this website?
Practical guidance for national statistical institutes (NSIs) on how to publish better statistics data. Covering data formats, file types, filenames, spreadsheets, methodologies and others hands-on aspects of data publishing.

What is this part of?
This is part of a three year project funded by Google as part of the AI Impact Challenge. It is a collaboration between Africa Check, Chequeado, Full Fact and the Open Data Institute to use technology to improve fact checking.

What are the project aims?

  1. To make the process of deciding what claims fact checkers check each day easier and faster.
  2. To help identify repeats of claims that fact checkers have previously checked.
  3. To make the process of fact checking be as fast as possible.

Where do national statistical institutes fit in?
To make the process of fact checking as fast as possible, fact checkers need access to good information. One of the most important of these is official statistics. Improving the quality of statistical publishing is a key factor in improving the speed and quality of fact checks and essential for any future forms of technical automation within fact checking.

This seems like a very big problem. Are you sure?
We are not looking to fix every part of this process, but we do want to make sure that we can explain how every office can improve in small ways that have a big impact on fact checkers. Most of this is focused on the way the statistics are published and in what formats. At times this will also involve slightly wider considerations such as underlying methodologies and how they are made available. Our intent is to not define a single solution for the ‘best’ publication, but working to ensure we have clear practical guidance of a series of small changes that are possible.

What is this guidance based on?
We base this guidance on a number of sources.

  • We’ve interviewed fact checkers in many countries around the world asking them about national statistics data they use and what they would like to see. Their feedback leads our guidance.
  • Much of the Open Data Institute’s guidance on data publishing.
  • Technologists on the project who have many years of publishing data sets, including for national statistical institutes.
  • Established standards and best practices for data publishing.

Which NSIs are you working with?
We are speaking with the Office for National Statistics in the UK (one of our project members was previously a deputy director there) and hopefully offices across Africa and South America Asia and North America.

What do you need NSIs to do?
To start with, we just need to speak to people in a number of NSIs to help understand the context they work in and why they publish content as they currently do. Over time, as we develop a clearer understanding of the change we would like to see across the statistical system, we would like senior members of staff in these organisations to work with us to promote our change agenda and champion its work.

Does this fit into an existing UN/EuroStat/ArabDevelopment programme?
At this stage, no. We are approaching this as fact checkers and the things we need to improve our work. We are happy to work with other organising bodies and programmes if this is appropriate, but at the moment can not see a clear match for this work elsewhere.

Is this just for NSIs with lots of resources?
No. Our aim is for this guidance to be helpful to all NSIs in different parts of the world with different levels of resources. While every organisation needs to deal with the fundamentals like filenames and contact information. Some organisations may reasonably choose not to use more advanced techniques like APIs.

Also, we realise that local context, language and practices are extremely important. We don’t provide this guidance as commandments that must be followed but more as advice that can be used when implementing your own data publishing standards.