Data science for the public good

My name is Jake, I’m a trainee data science lecturer at the Data Science Campus and I’m passionate about making sure our data science training offer is as accessible as it can be. I’m proud to share that we have implemented a new accessibility process in our course materials and training. In this blog post, I will talk about the process I developed and why this project is important, not only to the Campus, but to me personally. 

A large-scale accessibility audit of many courses and presentation templates on the Learning Hub and the ONS website in 2021 revealed many accessibility issues that would impact users of screen reader technology.

Following the audit, Richard Leyshon, a former data science lecturer in the Campus faculty, began developing the accessrmd package, now published on the Comprehensive R Archive Network (CRAN). It can be applied to courses written in R Markdown, where it is used to mitigate accessibility issues before exporting them to HTML. It also corrects many issues that would be caused by the HTML course structure, such as missing alternative text for images, as well as issues with contrast and reading order, which conflict with screen reader technology used by many people in our training programmes. 

The package improves the accessibility of the content we produce and has had a big impact, not only on my team and our customers, but also the wider public sector. 

The problem

Before the 2021 audit, our Python content was produced in Jupyter Notebooks and exported as a HTML. This was saved as course notes we would share alongside the ‘raw’ Jupyter Notebook files (known as ipynb files, or I python notebook files).  

Unfortunately, the audit revealed issues with the HTMLs produced. When running this process in RMarkdown, the issues didn’t exist (or were much easier to fix). R Markdown allows you to create Python content and export that to HTML, so this change was implemented immediately. 

No solution is perfect and there are still small issues that need to be addressed such as alternative text for images, contrast and colouring. 

The accessrmd package

The issues found in the RMarkdown HTML files prompted development of the accessrmd package to address them. Its functions will search for accessibility issues in your R Markdown and attempt to fix them so you can then re-render and produce a more accessible HTML file.

The package can be developed further to solve some more niche issues and this work is ongoing.  

The package is still useful for resolving many issues now and has been published to ensure our materials are as accessible and inclusive as they can be now. 

The package is the centrepiece of the process I have worked to develop with the faculty team and disability groups and there is now a solution that is applicable for documentation in both languages. Whilst it is incredibly useful to have accessrmd to mitigate many issues with our documentation, we believe that documenting the errors themselves before and after its use will establish a paper trail for any accessibility champion to pick up going forward, as well as contribute to the updating of the package in the future.

Guidance is available for using the Wave Webaim tool to document the accessibility errors before applying the package and keeping a record of those that remain. These can be added to the documentation of the accessrmd package to be worked on going forward.

Creating accessible documents in Microsoft Office

The Microsoft Office Suite is also important in developing training materials. For example, most of our show and tell meetings share slideshows and accompanying Word documents that are inaccessible. 

Following some investigation, I discovered the incredibly useful (but often overlooked) Review tab in the ribbon, which features a Check Accessibility option. The feature is very in depth, pointing you to the issues throughout your documentation or presentation and even providing tools to fix them in the same tab. I have included links to these pages in the process document to help users make accessible documentation at every level, not just for technical training.

This is a first big step towards our commitment to accessibility and I hope it inspires you to better your practices not only at the individual level, but also across your entire team, so that you can have accessible HTML files, PowerPoint templates and Word documents.

The importance of this work

Changes like these have been a long time coming. For too long many have been excluded from the conversation and we want to do all we can to make sure Diversity and Inclusion are at the heart of everything we do.  

Whilst this is just the first step to tackling this issue across the Public Sector as a whole, I believe it is very important to highlight the process we are implementing.

I hope you will be enlightened in the same way my colleagues and I have been in recognising the plight of those unable to engage with our materials and the tools to help them.  

If you want more information or need any guidance, please contact the Data Science Campus faculty. Data.Science.Campus.Faculty@ons.gov.uk