News & Events
Last month our Innovation Lab collaborated with one of our data engineering teams to conduct a research project which discovers the most effective ways to find and recruit diverse data talent.
The team uncovered five key findings and created a tool which allows organisations to benchmark their workforce against UK diversity metrics. We will be posting the detailed findings from the study here over the next six weeks.
Visualisation: pool of STEM/data-aligned graduates vs degrees required in job ads (click on image to enlarge)
3. Expand your skills search
In the next part of our study, we looked at over 10,000 job posts for UK data professionals with job titles ranging from data analysts to software engineers to understand what blend of hard and soft skills employers were recruiting for. We noticed something interesting.
We found the top five most required soft skills listed in the 10,000 job adverts were: communication, partnership, passion, flexibility and analytical. Both the top five, and wider soft skills required, were largely consistent across each of the role types in the data job family. In contrast, the technical skills varied substantially between job role, as you would expect. Notably, SQL, Excel, C, Python and Git remained core to each of the technical skills requirements.
Visualisation: top soft skills vs top hard skills (click on image to enlarge)
Industry commentary and feedback from our clients continues to highlight the importance of soft skills in data professionals and the ability to work effectively within a business unit. We believe the right approach is to hire for these attributes which are consistently sought after and where required, train the technical skills as the needs of the market evolve. Technical excellence in data can only be fully harnessed when it’s fully integrated into an organisation.
In our first post we revealed that over 30% of data professionals working in the UK didn’t study a traditional STEM subject at university, choosing instead a ‘data-aligned’ subject such as economics. We wanted to discover what impact this insight would have on the data talent pipeline and improving gender diversity.
We analysed the data talent pipeline through the education pathway of 173,000 UK graduates from 2018. We were particularly interested in the gender balance as students began to select their academic specialisms and understand how this presented challenges to recruiting more women into data.
It is widely reported that too few women are choosing to study STEM subjects in the UK, and the subsequent implications this has on the talent pipeline for UK industry, including data. We wanted to quantify this pipeline and understand how making changes to job application requirements can help unlock the potential data talent of the future who are studying data-aligned subjects.
We built a tool (see visualisations 1 & 2 below) to visualise the education pathway of UK graduates to then filter on different requirements: top universities, degree classification, STEM vs. Data-aligned etc. Our study found that by retaining a focus on high calibre graduates from both STEM and ‘data-aligned’ academic disciplines it is possible to almost triple the size of your applicant pool (24,000 to 62,000) and more than double the proportion of women within that pool (28% female to 57% female). This remarkable increase in both the size of the applicant pool and the huge increase in the proportion of women is a result of negating the gender imbalance in STEM subjects at undergraduate level by widening the search for talent.
Visualisation #1: STEM subjects, 1st & 2:1 degree classifications (click on image to enlarge)
Our study has shown that by recognising the potential for data talent beyond STEM it is possible to overcome this structural gender imbalance in higher education. By recruiting more women into data, as an industry, we will pave the way for future generations of young women to follow.
5. Watch your language when searching for data talent
Having explored the challenges and potential solutions to recruiting more women into data, we turned our study to examine the way recruiters were describing data roles in over 10,000 active job posts.
We built a natural language processing tool to review the 10,000 job posts and analyse the types of words used to describe six of the most common data roles. Drawing on published social science research, we tailored the tool to look for words which are characterised as ‘masculine-coded’. These words often constitute language which are assessed to discourage women from applying to roles due to their subtext.
Having analysed the adverts for data architect, data scientist, data analyst, data engineer, developer and software engineer, we found that over 70% of the adverts in our study contained masculine-coded language, as high as 81% for data architects job roles. What became clear from this part of our study was firstly, how masculine the descriptions of data roles tend to be, and secondly, how this is an easy win for organisations seeking to create the broadest appeal to job applicants.
Visualisation: Proportion of job posts with male gendered language by role (click on image to enlarge)
It is worth noting that the social science research, which formed the basis of the natural language processing analysis, found that a shift to more neutral or even ‘feminine-coded’ language had no discernible impact on the number of male applicants for roles while increasing the number of female applicants.
Posted on August 12, 2019