• Home
  • >
  • Editorial
  • >
  • DataIQ Leaders briefing – Creating a diverse and inclusive data office

DataIQ Leaders briefing – Creating a diverse and inclusive data office

In the first of DataIQ’s "Leaders-led" roundtables, members discussed the importance of diversity and inclusion within their organisations, shared best practice examples and looked at how the data industry can come together to drive change. The roundtable was part one of DataIQ’s "Talent in Data" series - a stream of talent-focused activities and content focused on the evergreen issues of talent retention, career development and diversity within data.
talent-in-data-retaining-talent-in-the-hottest-of-hot-markets

Diversity in the hiring process

Any journey to create a more diverse data office begins at the hiring stage. As organisations seek to address disparities in racial, social and neurological representation, boards are increasingly turning to targets in workforce headcount to help map out and drive diversification. However, many data leaders will tell you that they simply aren’t getting a diverse enough pool of applicants to hit those targets. Broader, systemic issues are discouraging talent from non-traditional backgrounds from applying in the first place.

Take job advertisements, for example. The typical listing for a data role will place an emphasis on technical skills. Indeed, a cursory glance at a LinkedIn job advertisement for a data analyst position lists within its essential criteria: experience with statistical analysis and experimentation; analytical mind and problem-solving aptitude; advanced SQL. These skills are undoubtedly important, but when technical language dominates a listing it can serve to dissuade practitioners from broader demographic or commercial backgrounds from applying.

Words such as analyst and analytics have been shown to appeal more strongly to men than women. By paying attention to the language used in job advertisements, for example by using a gender decoder, data leaders can build inclusion into the hiring process right from the start.

“Systemic issues are discouraging talent from non-traditional backgrounds from applying in the first place.”

The same consideration should be given to the application process. Traditional written applications can place neurodiverse applicants at an immediate disadvantage. One member from a large media organisation has sought to make this process more inclusive by broadening the mediums through which people can apply. They said: “We had some feedback from a dyslexic candidate, which we’ve used to inform our new approach to allow for audio and visual statements.”

Where possible, any panel of interviewers should reflect the range of diversity it is seeking to bring into the organisation. “Pay attention to the people you include in the interview process,” said one member. “We recently had a round of interviews, and at each stage I ensured we had a diverse panel on all fronts. This was quite successful, and we hired two great candidates from very different backgrounds.”

Results won’t be perfect – equality of opportunity does not automatically lead to equality of outcome. However, by paying attention to the recruitment process data leaders can ensure that all applicants are given a fair chance.

Example: Photographic evidence of bias on Twitter

Twitter’s automatic image cropping feature had been in place for years before anybody spotted a problem, but in September 2020 PhD student Colin Madland noticed something odd. When he uploaded a photo of himself and a black colleague onto his feed, the cropped image automatically centred on Madland, who is white. Further experiments revealed that this flaw was consistent, and the set of algorithmic tools that informed the technology had an inbuilt bias towards white faces.

In a statement, a Twitter spokesperson said: “Our team did test for bias before shipping the model and did not find evidence of racial or gender bias in our testing. But it’s clear from these examples that we’ve got more analysis to do. We’ll continue to share what we learn, what actions we take, and will open source our analysis so others can review and replicate.”

The social media giant has since phased out the old, biased system, but the lesson is clear: no data will ever truly be free of bias. When testing an algorithm – as Twitter claims to have done – it is therefore possible for the test data itself to be inherently biased. Unless challenging views are presented from a diverse data team, both the test and the product will inevitably carry forward those built-in flaws.

 

Promoting diversity of thought

Maths and physics grads have been the traditional lifeblood of the data office and with good reason. The core tasks of any data team require the analytical mindset and problem-solving skills that are typically found among STEM graduates. However, as the role of data office in steering broader business strategy grows, commercial competency is becoming increasingly important.

“It’s more about diversity of thoughts and opinions now,” said one member from a large retailer. “Maths people think in very black and white terms, and we need people who can think in a more commercial way in order to solve problems differently.” To bridge this divide, data leaders are looking to appeal to candidates with commercially awareness. As one member put it: “No senior executive really cares about the algorithm, they just want to know how they’re going to use it to increase profitability.”

In recent years organisations have hired “translators” to sit between the data team and commercially-focused business units. Another potential fix is to bring commercially-oriented professionals from the wider business into the data team. However, roundtable attendees agreed that opportunities within the data office aren’t particularly well known, or well publicised, to non-traditional candidates.

“Data science didn’t really exist 10 years ago,” said one member. “We need to show people what a career in data science can look like.” Another member from a large pharmaceutical organisation commented that they had worked in the business for years without knowing it had an insights team. “If I had known it was a potential career I’d have moved across sooner,” they said.

“It’s on leaders to identify and develop those with the drive and potential.”

On the other hand, data science practitioners with traditional backgrounds in STEM can only shake off their reputation for lacking business-savvy if given the opportunity to do so. “Traditionally, data scientists haven’t been given sufficient opportunity to speak to the business,” said one member from a large retailer. “Hiring a translator is a temporary fix, but data scientists should be encouraged to have meaningful conversations with the business to break down those barriers and open their eyes to the organisation. I think we’ve tended to put them into the wrong box.”

That said, not all data scientists will want to move out of their comfort zone. It’s on leaders to identify and develop those with the drive and the potential to move into the commercial space. One member said: “It’s about giving people the opportunity to step up but giving them a degree of protection so that if it isn’t quite right, they can migrate back into what they are comfortable with.”

Creating an inclusive environment

In November 2020, British director Tristram Shapeero had to publicly apologise after leaving himself unmuted on Zoom while criticising an auditioning actor’s “tiny apartment.” The unfortunate episode highlights one of the many by-products of lockdown: the blurring of the lines between home and working life. This has had some positive benefits, namely in the personalisaton of working relationships, but it has also shone a light on the need to create working environments that cater to different needs and backgrounds.

“It takes a long time to understand the factors that influence an individual,” said one member from the entertainment sector. “Education, first language and housing situations are all important factors that influence how certain practitioners work.” To some extent the office had been the great leveller, but hybrid working has brought new challenges for young parents and young professionals working in rental accommodation. “It is increasingly important that you empathise with each individual and create an environment that enables them to succeed,” they said.

“Race, gender and neurological diversity all play a role in our environment.”

Race, gender and neurological diversity all play a role in our environment, both in terms of comfort and productivity. An inclusive environment will support the diversity of experience and thought needed for the modern data office to thrive. “I do my best to identify challenges and encourage communication about what practitioners are comfortable with,” said one member. “By doing so I’ve created an environment where I’m more likely to retain people, while also getting the best out of them in the situations they’re really good at.”

As shown by the ingrained bias in Twitter’s algorithm, data and analytical models are influenced by the humans creating them. By cultivating an environment that supports diversity, data leaders can ensure that the biases inevitably built into their outputs aren’t skewed towards a particular demographic. Any data model is only as good as the data supporting it. Any data office is only as good as the experiences, backgrounds and thought processes driving decision-making.

Key takeaways

  • Level the playing field – Pay attention to the language used in job advertisements. Certain terms are off-putting to particular demographics, and traditional application methods might not play to everyone’s strengths.
  • Broaden your thinking – The traditional data office has been criticised for lacking commercial awareness. This could be overcome by making data roles accessible to a wider audience, both internally and externally. Data practitioners should also be given the chance to develop their business acumen.
  • Humanise the workplace Everyone is guided by a set of personal experiences and influences. By considering these, alongside styles of working, data leaders can create a working environment that supports and ultimately gets the best out of everyone on the team.

Upcoming Events

No event found!
Prev Next