A dangerous bias
Artwork by Valeria Fernandez-Soriano
Matthew Bazley explores how under-representation in STEM can have huge consequences for our daily lives.
In February 2019, UCL backed a proposal for an inquiry into diversity in STEM education. But is increasing diversity within STEM good for society overall?
Group dynamics are a major factor in research effectiveness. Most scientific advances are not achieved by individuals, but teams. In a 2010 study, participants were given a wide variety of tasks and puzzles to tackle in groups - including those requiring moral judgement and team decision making. Greater gender diversity led to better social perceptiveness, equality in participation and, ultimately, better performance.
Under-representation of minority groups in STEM research also affects technological and scientific advancement. Lack of diversity amongst workers leads to bias, often with disproportionate effect on minority groups and women.
For example, in biomedical research, despite significant biological differences between the sexes, women are still under-represented in research for certain health issues. Mainly cardiovascular disease, but also HIV, kidney disease, and digestive diseases. This even holds true in animal studies: in 2015, just 21% of all rodent-based papers in the journal Pain used both males and females, despite women making up 70% of pain patients.
Male rodents are regarded as more reliable models because female rodents, like their human counterparts, undergo a menstrual cycle which has wide ranging biological effects. This would mean more tests to accommodate hormonal fluctuations and sex differences. These take time and money that many researchers don’t have. Diversity in research complicates the results, precisely because it yields a fuller picture – good reason in itself to use diverse samples.
Widening the research pool can have benefits for us all, by improving the overall understanding of a disease. For example, by including more women in heart disease research, we have found new diagnostic techniques for women – because the symptoms and pathology in women are different to those in men – and have also challenged the long-held idea of the cardio-protective effects of oestrogen. Studying sex differences of brains in animal models has led to new treatments for Traumatic Brain Injury (TBI), which is more common in men than women. Research into the hormonal effects of menstrual cycles in women has even led to better understanding of auto-immune disorders.
In 2016, new regulations by the NIH require scientists to account for sex as a biological variable after it was discovered that a significant number of women were overdosing on the sleep aid Ambien. It was not tested on women, overlooking the crucial fact that men and women metabolise certain drugs differently. The dosage for women has now been cut in half.
We live in a world mostly designed by men and for men by default. For products that try to meet demand from both sexes, it is in the designers’ financial interests to cater to both. In 2014, Apple released its Health app, a tracker using the iPhone or Apple Watch. One of the main selling points was the sheer variety of biometrics: anything from blood alcohol level to mineral intake to respiratory rate. Apple, a male-dominated company, left out period tracking (for which there were already apps like Clue on the market), thereby missing crucial appeal to half of its potential market share before it had even released.
Sometimes, the lack of diverse representation in design can have fatal consequences. In a car crash, women are 47% more likely to be seriously injured. Due to the average height difference between men and women, women tend to sit forward when driving to reach the pedals, in a more compromised position. Women also have, on average, less muscle in the neck region, increasing the risk of whiplash. Issues like these could be avoided if cars and crash-test dummies were designed with diversity in mind. The US only began using female crash-test dummies in 2011. Even so, they are just scaled down versions of their male counterparts (disregarding body proportions) and are only placed in the passenger seat – a perfect metaphor for the problem.
The social impact of improving diversity in STEM is less easy to quantify, but perhaps even more significant. If you were to ask someone to picture a scientist, they would almost invariably imagine a white male, an Albert Einstein type. This unconscious bias against minority groups puts them at a disadvantage from even the earliest years of child development and education. Improving STEM diversity would aid in both eliminating harmful stereotypes and inspiring future generations of scientists.
As we move into the future, we must be aware that machine learning and AI are likely to capture existing biases in ethnicity, religion, and gender. Already, Google Translate massively overuses male pronouns even where the text is referring to a woman.
Perhaps if women and minority groups were properly represented in positions to make decisions within STEM fields, humanity’s needs could be better addressed. As humans, our ability to solve complex problems is limited by our narrow individual perspectives. The benefits of true diversity in a sector as economically and socially important as STEM are for humanity as a whole.