Skip to main content

Verified by Psychology Today


Overcoming Bias and Discrimination: Structure Over Training

Harnessing structured approaches to eliminate discrimination in selection.

Key points

  • Contrary to its popularity, research shows that anti-bias training is ineffective in fighting discrimination.
  • It's essential to prioritize structure over training, as it is shown to bring more validity and fairness.
  • New research shows that AI could be an effective way to standardize decisions and reduce discrimination.

The popularity of anti-bias training is on the rise, as seen in initiatives like the proposal from the California reparations committee for mandatory training for medical school graduation. Likewise, in France, legislation enacted in January 2017 requires companies with a minimum workforce of 300, as well as recruitment firms, to provide non-discrimination training to employees involved in candidate selection. While some individuals perceive this training as an effective approach to combating hiring discrimination, others consider it inadequate or meaningless. Moreover, there is a concern that such training might unintentionally amplify unconscious biases instead of alleviating them. Rather than relying solely on training programs and awareness campaigns to alter mindsets, it is crucial to prioritize the structuring of hiring processes that proactively prevent discrimination.

Is anti-bias training nonsense?

Numerous studies demonstrate the ineffectiveness of anti-bias training in producing lasting behavioral change. In essence, research clearly shows that diversity training does not work to reduce bias over time, nor does it increase diversity. In a recent experiment studying the effects of cognitive training interventions on reducing hiring discrimination against ethnic minority job applicants in the resume-screening stage, although discrimination was reduced shortly after the training, it resurfaced three months later. Being aware of social stereotypes and unconscious biases does not guarantee sustained behavioral modification. The human brain, shaped by a lifetime of experiences, is naturally susceptible to cognitive biases that unconsciously influence decision-making. The idea of making purely objective choices based on personal evaluation is a utopian ideal, as hidden reasons underlie all decisions. Short-format training courses lack the necessary structure to effectively mitigate cognitive biases or bring about enduring change. Additionally, it results that unless everyday discriminatory acts can be addressed, there is little usage in adopting unconscious bias training in the workplace. Furthermore, training that claims to limit unconscious bias could, ironically, contribute to activating stereotypes. According to Daniel Wegner, a former Harvard psychology professor, this rebound effect suggests that deliberate efforts to suppress thoughts often lead to their resurgence. Try to not think about a white bear, and you’ll inevitably end up thinking about a white bear. For example, individuals who undergo training to suppress age-related thoughts often display a tendency to evaluate older candidates unfavorably, whereas trying to ignore identity factors like ethnic origin can increase discrimination against Black applicants. Thus, relying solely on intense training without a structured and consistent approach is misguided in combating workplace discrimination.

Bring more structure

Recruiters often overly rely on intuition and resist analytical approaches when evaluating candidates. This subjective view of the assessment process neglects its probabilistic nature. To address this issue, it is crucial to introduce structure into recruitment processes by utilizing objective data collected consistently. This approach is essential for avoiding discriminatory decision-making. Renowned psychologist and Nobel laureate in economics, Daniel Kahneman, highlights our tendency to place excessive confidence in our own impressions and judgments. To mitigate this bias, introducing more structure in candidate selection, such as utilizing computer evaluation tools, can reduce discrimination and enhance the alignment between candidate profiles and job requirements. Similarly, conducting structured interviews improves the validity and fairness of assessments. Although the implementation of structured interviews may necessitate additional resources and time, they provide the advantage of being grounded in a comprehensive analysis of the position and enable evaluation based on consistent and standardized criteria. For instance, a recent meta-analysis of validity estimates in personnel selection found that structured interviews emerged as the most effective selection procedure, considering both validity estimates and diversity trade-offs, particularly regarding differences between Black and white subgroups.

Another potential solution for structuring hiring processes and mitigating biases is the adoption of hiring algorithms, enabling us to surpass our intuition and cognitive biases, by bringing standardisation to hiring decisions. Recently, scholars have been advocating for the use of such algorithms to reduce implicit biases in hiring processes and have proposed frameworks to evaluate AI-assisted interventions. By using algorithms for hiring purposes, employers will be able to control for not only gender bias but other discriminatory characteristics as these technologies are able to be trained in a way to filter through the necessary characteristics required for candidates and to ignore other features. This is supported by recent work showing that AI is equal to or better than human recruiters when it comes to efficiency and performance and that AI is mostly better than humans in improving diversity. For example, some algorithms, based on supervised learning and Upper Confidence Bounds (UCB), or also personality-based hiring algorithms, could increase the share of women selected, up to a balance of 50%, compared to 35% for hiring decisions made by humans. Hiring algorithms could also benefit by increasing the perceived equity of the hiring process: indeed, women prefer to be judged by a hiring algorithm because of its perceived objectivity over a human evaluator, while people are less offended by algorithm-led discrimination.

Training aimed at addressing discrimination, particularly when it focuses on behavior change rather than skill acquisition, often proves ineffective despite good intentions. This is evident in the persistent lack of progress in reducing discrimination against African-American communities in the United States over the past three decades, or in changing stereotypes and implicit theories about leadership. Many of these training programs fall short because their intention to deconstruct cognitive biases and alter thought patterns inadvertently activates stereotypes instead. Given the magnitude of the challenge, it becomes crucial to prioritize active participation in structuring the recruitment process using suitable, structured, and standardized tools and processes, rather than relying solely on training initiatives.

More from Emeric Kubiak
More from Psychology Today
More from Emeric Kubiak
More from Psychology Today