Clive Humby, one of the brains behind Tesco Clubcards, said “Data is the new oil,” by which he meant that there is commercial value in the exploitation of the data that is collected by business. David McCandless, author of Information is Beautiful, said in response, “Data is the new oil? No: Data is the new soil”, by which I think he meant that the exploitation of data was fertile ground for the imagination and understanding the world and not just for making money.
This is the first of PHE’s new series of data blogs. In these blog posts we take our inspiration from the Guardian Data Blogs which have been running for some time, Edward Tufte, and the burgeoning practice of data visualisation or vizzing as it is known. There are other inspirations, like Gapminder and the work of Hans Rosling, and we noticed this blog post in the BMJ yesterday. As well, according to Tableau, four of the top five visualisations of all time are about public health issues (including John Snow’s cholera maps, Florence Nightingale’s coxcomb diagrams, Joseph Minard’s map of Napoleon’s march into Russia and Gapminder).
We hope to take a population health issue and encapsulate it in an image or visualisation to try and make it easier to understand or to provoke comment and discussion. Hopefully the visualisation will help clarify what the data means or make it easier for people to explore and extract meaning and will not obfuscate or be gimmicky. We are visual animals and we remember things better as pictures than words or sound. A good visualisation will simplify and clarify a mass of otherwise complex information. It will be the “soil” from which we can enrich our understanding.
For the first of these posts I want to highlight this infographic from the Centre for Diet and Activity Research in Cambridge which summarises an analysis of the National Diet and Nutrition Survey (click through to see the full-sized version).
It is a “bubble chart” which shows how food group consumption varies by educational attainment and income in the UK. Each of these is stratified into three groups, subdividing the data into a three by three grid. Data is presented for 11 distinct food groups. Three things to note are that:
- the position of the bubbles has no intrinsic meaning but the position of each food group is the same in each square.
- data is only presented where food is consumed in statistically significantly greater quantity than the UK average
- the area of each bubble is proportional to relative excess consumption.
So what does it show? It summarises our diet based on food groups into a single figure. There is a lot going on but if we, for example, focus on the lower income and lower educational attainment square in the bottom left hand corner, there seem to be fewer, smaller bubbles than the top right hand corner representing higher income, higher educational attainment groups. That suggests a more varied diet. Other findings include:
- more fruit and vegetable consumption (green blobs), in higher income groups than in lower.
- more fat consumption and more “unhealthy” fat consumption in lower income and educational attainment groups (more transfats, chips and butter).
- more fish consumption in high risk groups.
The authors have made life a little easier by publishing similar grids for each food group separately. (They have also published an interactive version for those with up to date browsers). The figure on the left shows clearly that fruit and vegetable consumption is related to income and educational attainment. Similar patterns are found for consumption of grains (the figure on the right) and nuts.
It isn’t easy to tell the effects of income and education separately but, for example, fruit and veg consumption patterns seem to be similar in people with degree education regardless of income. Consumption of white bread is higher in people with up to GCSE levels of education. For some reason higher levels of consumption of game birds seems to be prevalent in people with degrees. These suggest that education may be more important than income per se in determining dietary patterns. Or it may be that levels of income and educational attainment interact to produce dietary patterns.
These kinds of display can’t necessarily give answers but they can generate questions and make us think, and because they put all the data in one place they paint the whole picture.
What can you see?
If you know of good visualisations of complex health data, or have done one yourself please let us know.
Featured image via Getty. Used under license.