I have a new article out in the journal Health & Place entitled OpenStreetMap data for alcohol research: Reliability assessment and quality indicators, written in conjunction with a number of people here at the OII and elsewhere. My colleague David Humphreys at SPI got me interested in the area when he told me about how difficult it was to construct local area indicators of alcohol availability in the UK, and how this was hampering research in the field. I wanted to see whether data in OpenStreetMap could fix the problem, as in general I’m pretty interested in the extent to which web data can be used as a valid proxy measurement for real life quantities of interest. Stefano de Sabbata, Sumin Lee and Bharath Ganesh all contributed to the analysis.
We did a few different things in the article: we conducted a validation of a random sample of 2,000 licenses we knew to exist, we used OSM data to duplicate a previous study in the area (E.A. Richardson et al. 2015 Is local alcohol outlet density related to alcohol-related morbidity and mortality in Scottish cities? Health Place, 33, 172-180), and we used a technique developed by Stefano to measure the ‘quality’ of OSM data in a given area. We showed that OSM is about 50% complete in terms of the amount of data it contains (in the specific case of alcohol licenses), and also that we could use the quality indicators to find areas with more complete alcohol data.
Quality of OpenStreetMap Data in Britain
Alongside the article, we are also releasing more general estimates of alcohol outlet prevalence across Britain, which are drawn from OpenStreetMap. We thought they might be of use to other researchers working in the area of the spatial availability of alcohol. They are a simple count of alcohol points of sale within each postcode sector in the UK, according to the data in OSM (see the paper for details of how they were counted). We’re also releasing an accompanying quality metric with each postcode sector so researchers can determine how trusted the OSM data should be (again see the paper for details on how it is constructed). The spatial distribution of the quality metric in the UK is mapped above. Feel free to reach out to me if you have any questions!
Get the estimates themselves here: OSM GB Alcohol Outlet Counts and Quality Index
The full reference for the paper:
J Bright, S De Sabbata, S Lee, B Ganesh, DK Humphreys. 2018. OpenStreetMap data for alcohol research: Reliability assessment and quality indicators. Health & Place 50, 130-136
and another related paper using the same dataset:
J Bright, S De Sabbata, S Lee. 2018. Geodemographic biases in crowdsourced knowledge websites: Do neighbours fill in the blanks? GeoJournal, 83, 3, 427–440
This research was partially funded by a grant from the ESRC (Grant no. ES/M010058/1).