It seems like everything is about BIG data these days so I was very happy to see this blog post from Public Health England, heard through Petra Boynton. The post discusses how PHE is working to break down national data on child health into local – ward level – units, to help inform local councils and commissioners. One major concern that the post discusses is that once you start to disaggregate large data sets, it can become difficult to maintain confidentiality or anonymity because records on individuals or communities where there are major outliers will stick out once they’re linked to a specific ward or area. But the potential for using local data to improve local practice and provision is clear, and it is great to see.
Nonetheless, even though this project is about providing local data for local providers, is still essentially a big data strategy – because it is about taking a large national dataset and breaking it into smaller pieces. This has lots of value – particularly the ability for local decision makers to compare their performance with performance in other areas. But isn’t it slightly odd that the central database is providing local providers with data from their own areas – and some of which they will have been involved in collecting in the first place?
For all the potential of big data I sometimes worry that we might make the mistake of thinking that every single piece of data has to go big, or, by extension, that only data that can go big (quantitative; comparable; systematically collected) is worthwhile. This problem is particularly common in the overseas aid field. Aid-funded projects, when designing data systems, often end up focusing exclusively on collecting the data that needs to be reported back down the chain to the donor, who can then break it up, analyse it, and maybe after a few months send some insights or lessons on the ground back to the people delivering services. I’d like to see data thinking that focus not only on big, strategic data and using it to its full potential, but also on enabling local, messy, perhaps unstrategic but definitely practical data collection and use, and that enables better local accountability and feedback. “Small data” approaches that also recognise that contexts and results (the things that big data tries to synthesise) are constantly shifting and that the best place to see and respond to those shifts is locally.
On a related point I would have liked to see the PHE blog post talk not just about how decision makers and service providers can use data, but also about how local people can use it to take a bigger stake in their local services and how they are run. And local people tend to find it much easier to understand and interpret small data that is about the local situation than percentages based on broad-stroke indicators collated at the national level.