How Data Science Diversity Can Actually Create Better Products
- By Devin Partida
- October 27, 2022
In virtually every industry, companies focus heavily on their diversity, equity, and inclusion (DE&I) efforts. Organizations with diverse teams tend to have a competitive advantage, allowing them to expand globally, scale up and meet their business goals.
It’s no secret that high-tech companies struggle with achieving diversity. Compared to the private sector, tech organizations employ a larger share of whites, Asian Americans, and men while hiring a smaller percentage of Hispanics, Blacks, and women. The lack of a diverse tech workforce could have negative, unintended consequences, potentially hindering a company’s success.
Data science is one tech-oriented discipline experiencing rapid growth as the volume of sources and information generated in the business world accelerates and increases. As a result, companies increasingly rely on data scientists to interpret it, use best practices and provide valuable, actionable insights to enhance performance. However, this industry is facing issues with diversity.
Companies that increase diversity in data science reap several business benefits — a prominent example being better products and services — that contribute to their success. Here’s more about diversity in data science and how a team of diverse employees can affect product quality.
Diversity in Data Science lacks
Recent research shows that only 15% of all data scientists are women and less than 3% are women of color. Historically, companies, researchers, and other industry experts describe this lack of gender diversity as a “leaky pipeline” problem.
In simple terms, the leaky pipeline refers to the idea that between middle school and graduate school, women tend to “leak out” of STEM fields or those that lead to professions in or related to data science. It furthers the notion that women and members of certain, often underrepresented, demographics tend to pursue careers in non-STEM-related industries.
The leaky pipeline theory is often used to explain the lack of diversity in data science, especially at the executive level. According to some women with successful careers in the sector, it’s not a leaky pipeline causing a lack of diversity. Instead, these women believe the current pipeline is lined with harmful practices undermining the presence and successes of females and people of color.
Dr. Fatima Abu Salem is a computer science professor and data scientist in Beirut, Lebanon. She says she’s often experienced discrimination from male colleagues.
For example, male co-workers reviewing her work in data science have described her writing as “flowery” and “dramatic,” saying that her research articles are inappropriate for communicating data to the audience. In addition, Abu Salem feels that a lack of women in the workplace causes male employees to set work standards, and women are expected to abide by them.
The business case for Data Science diversity
Why should companies operating in the data science field make diversity a priority? The simple answer is that these businesses and employees can reap several benefits by focusing on building a more diverse workforce.
Aside from being a morally sound business practice, prioritizing diversity and inclusion can help companies:
• Improve decision-making
• Maintain compliance with applicable laws and regulations
• Offer better opportunities for professional development and growth
• Win over qualified candidates in the talent pool
• Retain high-performing employees and boost levels of employee satisfaction
• Increase productivity
• Stay ahead of the competition
In addition to the items on the list above, diverse workforces help companies create better product and service outcomes for their customers and clients.
Diversity and product outcomes
Companies at the forefront of the tech sector, including those offering products and services related to data science, benefit from the creativity, innovative attitudes, and out-of-the-box thinking that naturally flow within a diverse workforce.
In other words, when companies hire employees with different backgrounds and unique life experiences, they bring new perspectives to the table and stray from standard approaches to product and service development.
Every company is looking for the next big thing to sell. It could be a revolutionary AI-based data analytics solution or a new consulting service. When diverse employees work on developing these offerings, the team will make fact-based decisions, feel more engaged, drive innovation and ultimately create better products.
Data science companies selling high-quality products will notice increased sales, an improved customer experience (CX), and a positive company reputation.
A diverse workforce could also eliminate conscious and unconscious bias in products. For example, researchers at Columbia University conducted a study to determine how algorithmic discrimination occurs. Around 400 AI engineers were tasked with creating algorithms to make millions of predictions about a group of 20,000 people.
After the study, researchers concluded that the demographics of the 400 engineers played a role in producing biased predictions. It was found that everyone was more or less equally prejudiced regarding race, ethnicity, and gender.
However, researchers noted that prediction errors correlated within the demographic groups based on gender and ethnicity. Researchers also mentioned that homogenous groups of engineers with the same demographics had a higher chance of making a given prediction error. More diverse groups will reduce the likelihood of compounding bias in product development.
Improving Product Quality
Increasing diversity in data science won’t happen overnight. However, companies in the industry should avoid using the leaky pipeline metaphor as a scapegoat for the lack of inclusion.
Data science organizations with diverse employees will reap many benefits, such as gaining the ability to create better products. As a result, these companies could outperform their competitors with workers of similar demographics, backgrounds, and experiences.
Image credit: iStockphoto/wildpixel