For every UK Healthcare Tableau User Group #UKHealthTUG we try to set a data viz challenge; encouraging participants to share a visualisation based on a pre-determined theme or task. Our March 2018 data viz challenge supported Tableau's #VisualizeDiversity campaign and asked TUG members to visualise diversity in healthcare.
Before starting my submission I searched the web for some inspiration; I found myself getting so mad at one particular NHS infographic that I felt obliged to make it the focus of my viz and give it a makeover.
What was this vile, evil infographic that got me so riled?
The infographic in question was one created by NHS Employers about the ethnic profile of the NHS workforce in 2016. It compared the ethnic profile of the NHS workforce with the ethnic profile of England's working population and then split the NHS analysis down by key staff groups. My main gripes with the infographic were:
Why did they need to repeat the ethnic group colours by putting them in a box in the 2nd section when the 1st section already had each ethnic group coloured?
In the ethnic breakdown for Agenda for Change staff and associated bands why does the ordering of the ethnic groups change for each of the 4 charts?
Why do the people icons used for the Bands 8a-9 section bare no resemblance to the actual ethnic groups? For example the ethnic group 'Black or Black British' is depicted by a white face!
Whilst some of the graphics used for the medical and dental staff sections are quite cool and engaging (for example I love the syringe) why do the graphics in the Agenda for Change section constantly change in a way that adds no value to the data and actually makes it harder to compare ethnic groups? Surely 1 common type of graphic per section would have been better.
The top section of comparing two populations through opposite facing bars is a commonly used technique; however in this example the bars are embedded within a white banner and I think this makes it hard to compare one population against the other. In addition what is the sort order used for this section? It is not descending by one population or alphabetical for ethnic group. It makes no sense to use an order where by the 2nd ethnic group is not the 2nd highest in either of the populations being visualised!
At this point I feel I should say I am a big fan of infographics; when done right they can help achieve greater engagement with the audience, immediately achieving a sense of intrigue which results in the audience wanting to find out more. In addition, and probably the greatest benefit of an infographic, they can help simplify complex data. Unfortunately the NHS Employers example achieves none of these benefits!
My makeover
For my makeover I knew I wanted to try and achieve a visual representation of the workforce profiles, as opposed to presenting the data through a set of tables. Here are a few of the techniques I adopted in my makeover:
When comparing the NHS workforce with England's working population I utilised a technique that I learnt from Charlie Hutcheson; embedding the 2nd measure as a smaller bar within the 1st measure. This allows the audience to easily compare 1 population with the other.
For the staff group sections I utilised treemaps to compare the ethnic profiles; however I ensured all the treemaps were ordered by descending populations so that the largest ethnic groups were always on the left, moving through to the smallest in the bottom right.
I didn't repeat the colour legend for the ethnic groups within each section as the colours were prominent in the top section through the use of labelling the ethnic groups using the same colours as applied to each group in the 2 size bar chart and the treemaps.
Whilst I recognise my makeover is not an infographic I think it is easier to interact with and to compare ethnic groups across the various sections of the viz; what do you think? I would love to hear your feedback.
My finished makeover can be viewed via my Tableau Public profile:
https://public.tableau.com/profile/simon.beaumont#!/vizhome/UKHealthTUGVisualizeDiversityvizchallenge-NHSWorkforceEthnicityProfiles/UKHealthTUGVisualizeDiversityvizchallenge-NHSWorkforceEthnicityProfiles
Yep, that NHS Employers viz was truly awful. I remember at the time a lot of people being interested in the data behind it and inquiring about the possibility of replicating and benchmarking against it for local organisations.
I recall several periods of head-scratching trying to understand it's internal design logic and wondering why such poor design decisions had been made. Needless to say, after much frowning, we advised against direct replication!
Nice to see you putting it right.