Terrific Scientific Analysis
In BBC Education we are in such a privileged position to be able to help young people, so when we were set the exciting challenge to help turn around some of the statistics below, we jumped at the chance:
- Each year, the UK needs over 100,000 new graduates in Science to keep our industries thriving but 40,000 STEM (Science, Technology, Engineering, Maths) related jobs are left vacant annually
- Only 15% of young people aspire to become a scientist
- Girls in particular are not attracted to selecting science in their later academic years
- Teachers find it difficult to keep children engaged and enthusiastic when teaching science.
And so Terrific Scientific was born.
What is BBC Terrific Scientific?
Terrific Scientific is composed of 10 hands-on science investigations that form school lessons for 9–11 year olds. These investigations each have a different theme and each pose their own questions. We hope to gather the results for these investigations with the help of each participant. All the details for these investigations can be found here on the Terrific Scientific home page http://www.bbc.co.uk/terrificscientific.
By running these investigations across England, Wales, Scotland, Northern Ireland, Isle of Man and the Channel Islands we are aiming to collect data at a scale never seen before. We’re partnering with national organisations and Universities who will use this large data set and produce scientific reports.
As well as our partners producing large scale reports, here at the BBC we’re also using a selection of data from each investigation to display the headlines on an interactive map which is the heart of the Terrific Scientific community. This is an open resource which lets everybody access the scientific data gathered for the map, not only helping the children of today but for children in the future too.
We’re also providing opportunities for every household in the country to participate outside of school to emphasise that science can be a cool thing inside and outside of the classroom and is everywhere around us.
How is the map composed?
I’m going to focus on the map that was built (http://www.bbc.co.uk/terrificscientific/map), so I’ll start by giving you a summary of what the map offers:
- A base version of the map showing all of the schools that have registered to take part in Terrific Scientific
- Each investigation then has it’s own version of the map, these are used to show all of the investigation results that are uploaded by the teachers through the use of data visualisation.
- The map is composed of multiple zoom levels
- The schools and their data visualisation is clustered together until the final zoom level, the scale of clustering becomes more granular as you zoom
- Each school is represented by a school panel, collecting all of the classes at the school and its results for each investigation
- There is a search panel enabling people to dive straight into the school they’re looking for
Data driven decision making
The number of schools to accommodate
We needed to know how many schools we had to accommodate to reach our target audience across England, Wales, Scotland, Northern Ireland, Isle of Man and the Channel Islands.
Through the Freedom of Information Act we were able to request most school data from each nation’s Department of Education to use as test data. That’s approx. 20,000 primary schools!
This gave us the core school details and from this we were able to decide the most suitable level of detail to show on the map at each point such as search, map, school panel etc.
This data became vital test data throughout the creation of the map before we had any real schools registered.
The proximity of schools
We had to consider how close primary schools were when planning the zoom levels and clustering of schools. We also had to ensure that everybody would feel included regardless of location. For example people in very isolated locations still had to feel a part of the Terrific Scientific community.
It was apparent that we wouldn’t be able to show every school individually at the higher zoom levels because there would be too much overlapping and the map would become unusable.
This meant we would have to find the correct level of clustering at each zoom level to ensure the core information was displayed and readable. To do this we had to mock out different potential scenarios in the extreme areas using geohashing.
Schools that share postcodes
We position schools based on postcode, postcodes cover distances and not an exact location. This means schools are positioned centrally within the postcode area so schools that share the same postcode would be positioned in the same place and not exactly where they are in the real world. We had to understand how many schools this would impact (approx figures):
• 202 primary schools in England
• 59 primary schools in Wales
• 240 primary schools in Scotland
• 12 primary schools in Northern Ireland
• 0 primary schools in Channel Islands
As these schools all show on the map in the same location we realised this would need addressing by finding a way to accommodate them and make them all findable and accessible.
One sacrifice we had to make as a result was to not show a lot of detail in the map design. A school being located within a postcode area isn’t difficult to find when the map is representational. If we went to street level people looking for a particular school wouldn’t find it pinpointed where they expected.
Displaying investigation data
Initially we wanted to use a choropleth map, containing regions, counties and boroughs, similar to the below. We thought this might help children identify where they are on the map and that this could also be a method to display the investigation results.
We found that across England, Wales, Scotland, Northern Ireland, Isle of Man and the Channel Islands there were variations of borders based on politics, religion and administrative areas etc. We did not want to show any preference towards one over another one.
To follow this approach we would also have to extract additional data for each school i.e. North West, Cheshire, Cheshire East and we wanted to keep the work required by each teacher to a minimum to make the process as streamlined and easy as possible.
Some schools also sit across different borders. For example the postcode for “Ysgol Carreg Hirfaen” is in Lampeter but the Lampeter postcodes appear to straddle two counties. Children at this age are still learning where they live in the country and showing them in the incorrect country or region may have been confusing for them.
The use of choropleth also poses some accessibility challenges; it relies solely on using many different colours. To enable as many people as possible to use the map we wanted to use a variation of colour, shapes and sizes to represent the investigation data. This approach helps people with different vision needs by offering an alternative view of the data.
Instead we used iconography to display the investigation results and did not follow the choropleth approach.
Aiding navigation
Through user testing (we regularly visited primary schools or showed children prototypes in a lab environment to gather feedback) we realised our target audience would need help finding their location on the map. We did this through the use of landmarks and town names.
Using the school data we gathered from the Department of Education we reviewed how many primary schools were in each town. The towns with the most primary schools were shown on the higher zoom levels so that we reached higher user levels. For example:
- there are 59 towns with between 1152 and 51 primary schools in them so these are shown on zoom level 1
- there are 42 towns with either 15 or 14 primary schools in them so these are show on zoom level 3
Once we had these town labels on a test version of the map we tweaked them so that the distribution of towns were more even across the zoom levels and so rural and urban areas had similar levels of support.
We decided to launch the map without any town labels or landmarks on the final zoom level. The reason for this was to ensure that there wasn’t any clashing of information with the schools/investigation data which in turn could make the schools inaccessible.
Once we had the map live we were able to make a better judgement on whether this would cause any issues and felt that it would actually aid users by having them. To gauge the correct level of labelling we used existing mapping sites to judge the correct scale.
Schools that share the same or similar names
Approximately 15.5% of schools have a duplicated name (shared with at least 1 other school).
The most popular school name is St Joseph’s Catholic Primary School, this is shared by 57 schools. This is only based on the exact name in the Department of Education records, there are also similar versions of the same name too which aren’t included in this count! This could be schools that use Saint instead of St etc.
This meant that we had to ensure the right level of school information was displayed wherever the school name was shown so that users knew they had the correct school. For example within the search we display the road, town name and postcode and within the school panel we display the town name.
There are various similar school names. If these schools are only searchable in these exact formats it will be difficult for children (and even adults) to correctly input their correct school name to return a successful result. Searching is really key in our map as the children we’re aiming at have been brought up in a Google world where everything is just a search away.
We expect the school name to be the most popular search term when searching for a school as it’s unlikely children in our target audience age group will know the school postcode.
We’ve been able to apply a set of grouping terms for synonyms within the search i.e. ‘CofE’ and ‘Church of England’ or ‘St’ and ‘Saint’. We’ve also removed common words i.e. ‘Church’ This should provide some flexibility when searching for a school name.
The length of school and town names
There are several places we need to show the school and town names; we needed to know how much space was needed to accommodate these, especially across the different break points.
The school with the longest name has approximately 86 characters.
We decided to cap the character length for the school name to 65. There are around 417 schools with over 65 characters in their school name.
Considering that is out of approximately 22,000 it’s a small portion. The school name research was based on the information from the Department of Education and following some quick searches the schools that exceeded the 65 characters didn’t seem to use the full name on things like their website or school jumpers etc so we thought these 417 schools would be manageable on a case by case basis if there were any issues.
The longest town name containing a primary school has approximately 22 characters. We felt we could accommodate this
in all the places it had to be displayed.
Accessibility
There are not many maps used by the BBC so we spent a lot of time consulting various people within the BBC about what our core accessibility considerations should be to ensure this is a product that could be used by everybody. This includes the accessibility team, News, Microbit, Children’s, My BBC and Platforms.
We’ve thought about the many ways people may control and interact with the map, such as mouse, touch, keyboard or screen reader.
For users with low vision or colour blindness we’ve made sure the contrast between the colours will enable anyone to access all of our content.
We even have a user guide: http://www.bbc.co.uk/guides/z8tgdxs
Conclusion
When scoping out any project the audience should be at the heart of it. To understand your audience you need data. There are various ways in which data can be sought and this should never stop throughout a project lifespan.
