In February the good people at UCAS released their end of year report along with some lovely dashboards allowing people to look into the major trends around applications and accepts for the UK HE system. They also released their massive range of data sets which gave the VizHE folks something to do in lock down.

With their shiny dashboards UCAS had pretty much covered the basic summarisation of the data so we wanted to see what else we could extract that piqued our interest.

With the current world situation we talked through looking at subject trends from a position of what are the popular subjects and to what degree had they been influenced by the brilliant scientists and doctors who have been a large part of the collective effort in the past year or so.

We each followed separate lines of enquiries and came up with three fantastic vizzies – one using the Boston Matrix approach to review subject trends, one digging into the Whitty effect and one looking at the choices both within subject and ethnic group around clearing and adjustment.

The Boston matrix is a tool used to look at the portfolio of products to see how they are performing. It effectively places the products into four quadrants:

Star (High market growth, high market share)

Question Mark (High market growth, low market share)

Cash Cows (Low market growth, high market share)

Dogs (Low market growth, low market share)

The quadrants are created by plotting the change in market share against the market share for the current year based on the applications for each subject. It is possible to look at the data by different date ranges and by individual institution performance.

The visualisation is available here: https://public.tableau.com/views/UCASApplications-BostonMatrix/InstitutionBostonMatrix?:language=en&:display_count=y&:origin=viz_share_link

Jennie – The Chris Whitty Effect – Analysing the increase in UCAS applications for health and social care courses

I decided to focus on the increase in applications for health and social care courses coinciding with the COVID-19 pandemic, termed ‘The Chris Whitty Effect’ by WONKHE in their recent blog available here (Tracing the Chris Whitty effect in the 2020 end of cycle admissions subject data | Wonkhe).

Delving into the data more closely I could see straight away the increase in JACS subject area ‘Subjects allied to Medicine’ which has a big jump from 2019 to 2020 and it stands out amongst other JACS subject groups. The trend over time is also interesting for this subject; applications dropped in 2017 and 2018 coinciding with the change in removal of bursaries from 1^st August 2017 for students taking pre-registration nursing, midwifery and allied health courses. This year though, main scheme applications (the main UCAS application scheme whereby students can apply for up to five institutions or course) are up by almost 9%.  Medicine and Dentistry has also seen a 5% increase in applications, but by far it is Subjects allied to Medicine that has seen the biggest increase.  When looking at total accepted places, again it is these two subject areas that are ranked on top, with a 17% increase in accepted places for Subjects allied to Medicine and a 9% increase for Medicine and Dentistry.  Conversely Non-European Languages and Literature has seen the biggest drop, with a 15% decrease in main scheme applications and an 11% decrease in total accepts, so it would be interesting to delve further into this data in future analysis.

Looking specifically at Subjects allied to Medicine I wanted to focus on individual providers to see if an increase in applications is also reflected in an increase in accepts. As I expected, the majority of providers that saw an increase in applications from 2019 to 2020 also saw an increase in total accepts. Of interest though are those providers for which applications increased but accepts either stayed static or even decreased. I wondered if this was due to caps on the numbers of places available at these providers so I looked at applications and accepts across all JACS subject areas for these particular providers.  Although I couldn’t see any obvious pattern here, some of these providers did see an increase in applications and accepts for certain subjects (see University of Bath).  This could mean that caps on places for Subjects allied to Medicine might be the issue, however more in-depth knowledge of courses at the provider would be needed to know for sure.

This time I experimented with using a long form dashboard which I haven’t used before.  I usually like to see everything all in one go, however I do think it works and helps tell the story.  I also played with the golden yellow colour across the dashboard, I don’t think it looks too bad, but let me know what you think 😊. If I had more time I would have liked to repeat the bottom chart for all providers in scatter plot to see if this pulls through any patterns when comparing across different subject areas.

The visualisation is available here: The Chris Whitty Effect – Analysing the increase in UCAS applications for health and social care courses – Jennie Holland | Tableau Public

Sasha – Widening Participation – Exploring changes in relationships of ethnic groups in UCAS application stages

I set out to investigate the difference between applications and acceptances recorded for each acceptance route against measures of widening participation in HE. The first challenge was to combine the files: there was one file containing applications, another acceptances and a set for each of the WP metrics. Starting with the ethnic group data I noticed that there was a huge difference in numbers; far more acceptances than applications. The crux being the applications file only contained main scheme applications whereas acceptance also included others (see below). Not a big issue as you can use this field to join the data. However what you end up with is only three acceptance routes: Firm, Insurance and Other, and not many insights to draw.

Back to the drawing board, I looked at the initial question that interested me. I wanted to look at the differences in groups of people going to university at the opposite end of the scale acceptance scale: adjustment and clearing. I dropped the combined file altogether and focused solely on acceptances. The below DNA chart shows the few groups going for adjustment and their subject choice by year. It also compares the popularity of these subjects when it comes to clearing.

The visualisation is available here: TBC

The Office for Students started to provide updates of the general student numbers at providers under their regulatory control. This covers both traditional higher education providers as well as further education colleges and alternative providers.

The data is available here: https://www.officeforstudents.org.uk/data-and-analysis/student-number-data/get-the-current-student-numbers-data/

The visualising HE team decided to give it a crack as it is a simple data set and it provided a genuine whole sector view of data.

This as it turned out was not as straightforward as we had hoped. The simplicity of the data set made it challenging to find the interesting story within the data.

This gave us an opportunity to try out different approaches and pushing both technical and design skills to try and come up with something that is interesting.

Each member of the team outlines below what they have produced and what they have gained from the process.

Dave – Size and shape over the last 2 years

https://public.tableau.com/views/SimplestudentnumbersV3/VizHE-OfSStudentFTE?:language=en&:display_count=y&:origin=viz_share_link

I struggled to find an interesting narrative at first then as I looked at the changes between the years I noticed that although the numbers for further education colleges were smaller than the changes in higher education the proportions of the total was much more dramatic.

Once I found this I decided that I would create a visual that stepped through the data to draw out this interesting element. The long form story telling nature of the visualisation focusing on bar charts hopefully works well.

Jennie – Simple student numbers reporting

My visualisation simply looks at reporting student numbers at different levels, allowing the user to select a higher education provider and compare how the proportion of students at different levels compares with other institutions. Student numbers on the right-hand side of the chart are provided for context and both 18/19 and 17/18 figures can be shown.

https://public.tableau.com/profile/jennie.holland#!/vizhome/HigherEducationStudentNumbers/Dashboard1

This analysis focuses on higher education institutions.  Unsurprisingly at the majority of these institutions 100% of students are taking level 6 or above qualifications, however I feel as though it is more interesting to look at the institutions which offer levels 4 and 5 qualifications.  For example, on average only 1.5% of students across all university institutions are studying for level 5 qualifications, however almost 30% of University College Birmingham students are at level 5.

Whilst the dashboard isn’t too complicated, it nicely uses sets, set actions and animations to reorder the main chart based on the proportion of students at the level selected.  The stacked bars also sort dynamically, with the selected level always at the end of the bar to allow easy comparison between universities.

James – student population comparison

When I began exploring the data in Tableau I was struck by the size of some of the institutions. That gave me the idea to try to find out how many large institutions it takes to match the populations of the remaining institutions. I decided that two stacked bars next to each other could show this.

The data cover both Further and Higher Education institutions and different education levels. So I built the workbook to allow you to make comparisons within and between them. The most striking finding is that when looking at all levels combined for 18/19, it takes only the top 3 Higher Education institutions to match the population size of all the Further Education institutions.

https://public.tableau.com/profile/jdis#!/vizhome/VisualisingHE_StudentNumbers/StudentNumbersbyTypeandLevel

Building the workbook required me to solve some technical challenges. To make the stacked bars gain or lose institutions I made some plus or minus buttons. Clicking these triggers a parameter action increasing or decreasing the number in the parameter by one. The number of institutions in the right stacked bar is based on the rank of the sum of the students at that institution. If the rank is less than or equal to the value in the parameter then that institution appears on the right hand side (and disappears from the left hand side where appropriate).

To make the buttons I put institution on columns in a sheet in Tableau and put index() on detail in the marks card. I then filtered to only leave the institutions with an index() one above and one below the current value in the parameter. When you click a button the parameter is updated and the button sheet is filtered to the institutions above and below that rank.

Blog number 2 for the 2020 VizHE team which now includes the skills of Rhodri Rowlands and Jennie Holland both visualisation masters and data wrestlers extraordinaire.

The standard data returns that UK HE institutions have to complete is focused on the estate that University inhabits.  This data set is not always the most exciting (although VizHE have covered it before here and here) but with the current position the sector, country and world are in the estate and it’s make up suddenly becomes an important topic.

The initial discussion for the VizHE team was to look at to what extent social distancing would be possible on campuses.  In the data set from HESA we have students, staff as well as internal and external space for each institution so a couple of calcs and we would have the answer….

As we talked it through though we realised we were getting caught up in the jump to an overly simplistic view of a complex problem.  There are many factors that the metric does not take into account such as how the spaces are configured, what groups the students are in and also to what extent the data focuses on students and staff who would be on campus!

So changes in direction needed – luckily this data set had plenty of options.Within the data set there is a wealth of information around the environmental credentials of the HEIs.  From waste to energy to emissions to travel.It was in travel that Dave decided to explore and to see how the UK HE staff get to work.  In this Viz Dave has defined four categories:

• Great – where staff cycle or walk to work

• Not too bad – where staff get the bus, train or car share

• Bad – where staff travel via motor bike or single occupancy car journey

HEIs were then ranked by how great they were.  There were some unsurprising results with many Oxford and Cambridge near the top and a lot of London and big city HEIs doing well.  Within this though there was a contrast between how many were really great against how much they were just not too bad.Interactive viz – click here

Jennie’s visualisation is titled ‘Uni-cycling – which institution is #1 for cycle spaces?’, and looks at the total number of cycle spaces at each institution against the number of staff and students to build a cycle space ranking.  The visualisation is split into two halves.  The left-hand gears control the ranking and can be used to flick through the ranking list.  The right-hand gears are used to highlight institutions and look for patterns based on groupings such as Million Plus, Russell Group etc. (data sourced from learning-provider.data.ac.uk), as well as their geographical location.

Unsurprisingly Oxbridge do well, with approximately 5 people per cycle space on average (2018/19) compared to 29 people on average across all other institutions.

To expand the analysis further it would be interesting to source some additional datasets and see if there are differences in the number of cycle spaces based on institution type (city, campus etc.), as well as how well the surrounding area is equipped for cycling based on the number of cycle routes. 

View the viz here!

Hopefully that has shown that there are some interesting insights in the Estates data!

Let us know what you think

VizHE crew

Hey up – it has been a while since the Visualising HE crew have got together.  Since the last post Elena has moved to Sweden so we have roped in Sasha from the Information Lab to bring much needed energy and skills to the project.

As with the post last year we have picked up the Finance data to see what interesting insights we could identify.  The talented bunch at WonkHE have been pulling the Finance HESA data apart in recent weeks in particular looking at the potential impact the current situation may have on the sector and indeed individual institutions – if you have not read their stuff then you really should.

We each took different routes through the data which brought the opportunity to try out new chart types.

Adam and Sasha both looked at the income streams for Universities to see how diverse they are.  Adam wanted to be able to the see the whole sector whilst also being able to see the individual Universities within the sector.  His Viz also allows the user to change the sort order if there is one income stream you are more interested in.

Adam’s Viz

Link to interactive viz

Sasha wanted to focus on the individual University so you can see how diverse the income is and which of the income streams dominates by University.  The visualisation also has the ability to see the information for previous years as well.

Sasha’s Viz

Link to interactive viz

Dave decided to focus on the picture of the sector as a whole and using basic quadrant analysis spit out the Universities into what degree their income is made up of tuition fee income and out of that tuition fee income how much is from non UK or EU students.  The visualisation paints quite a stark picture given the current position.

Dave’s Viz

Link to interactive viz

Well that is the first post for Visualising HE for 2020 – let us know what you think