Countering Selection Bias

Countering Selection Bias

Regardless of the social groups we belong to, we all perceive people differently based on their demographic characteristics (race/ethnicity, gender, sexual orientation, gender identity, disability, religion, politics, etc.). However, and importantly, most people try to overcome their stereotypic preconceptions. In searches for academic personnel at UC Berkeley it is unacceptable to act on biases, conscious or unconscious. There are many successful strategies for overcoming the tendency we all share to fall back on preconceptions and stereotypes in decision-making.

Diversity offers significant advantages

Recent research reveals advantages of diverse groups in academia and industry. People who are different from one another bring unique information and experiences, and diversity promotes creativity.

One study found that female representation in top management leads to an increase of $42 million in firm value (Deszo and Ross, 2012). Another study found that papers written by diverse groups have more citations and higher impact factors (Freeman and Huang, 2014). And diverse groups also share more information, while being similar with others makes people believe they all have the same information (Neale, Northcraft, and Philips, 2006).

Assumptions influence the review process

We all like to think that we are objective scholars who judge people based entirely on their experience and achievements, but research on bias in selection shows that every one of us brings a lifetime of experience and cultural history that shapes the review and evaluation process.

The results from studies in which people were asked to make judgments about subjects demonstrate the potentially prejudicial nature of the many implicit assumptions we can make. Examples range from physical and social expectations or assumptions to those that have a clear connection to hiring, even for faculty positions. Consider taking the Implicit Association Test developed by researchers at Harvard to develop a better understanding of how implicit assumptions operate.

It is important to note that in most of these studies, the gender of the evaluator was not significant, indicating that both men and women share and apply the same assumptions about gender. Recognizing biases and other influences not related to the quality of candidates can help reduce their impact on your search and review of candidates.

Findings on bias in academic evaluations

  • Professors at top Universities were contacted by a fictional prospective graduate student. Faculty ignored requests from women and minorities at a significantly higher rate than requests from Caucasian males, particularly in higher-paying disciplines and private institutions (Milkman, Akinola, & Chugh, 2014).
  • Research participants redefined the job criteria as requiring credentials that matched those of the desired gender. Commitment to hiring criteria prior to disclosure of applicant gender eliminated discrimination (Uhlmann & Cohen, 2005).
  • A study of postdoctoral fellowships awarded by the Medical Research Council in Sweden, found that women candidates needed substantially more publications (the equivalent of 3 more papers in Nature or Science, or 20 more papers in specialty journals such as Infection and Immunity or Neuroscience) to achieve the same rating as men, unless they personally knew someone on the panel (Wenneras and Wold, 1997).
  • “Blind” auditions can explain 30% to 55% of the increase in women winning orchestral jobs (Goldin & Rouse, 2000).
  • A study of over 300 recommendation letters for medical faculty at a large American medical school in the 1990s found that letters for female applicants differed systematically from those for males. Letters written for women were shorter, provided “minimal assurance” rather than solid recommendation, raised more doubts, and portrayed women as students and teachers while portraying men as researchers and professionals. All letters studied were written for successful candidates only (Trix and Psenka, 2003).
  • Another study showed that the preference for males was greater when women represented a small proportion of the pool of candidates, as is typical in many academic fields (Heilman, 2001).
  • Evaluators who were busy, distracted by other tasks, and under time pressure gave women lower ratings than men for the same written evaluation of job performance. Sex bias decreased when they were able to give all their time and attention to their judgments, which rarely occurs in actual work settings (Martell, date).
  • When a male instructor mentioned a male or female partner, the “straight” instructor received 22% more positive comments, while the “gay” instructor received 320% more critical comments (Russ, Simonds, & Hunt, 2002).
  • In a national study, 238 academic psychologists (118 male, 120 female) evaluated a resume randomly assigned a male or a female name. Both male and female participants gave the male applicant better evaluations for teaching, research, and service experience and both were more likely to hire the male than the female applicant (Steinpreis, Anders, & Ritzke, 1999).

Mitigating the effects of bias

  • Be systematic about evaluation criteria – select them ahead of time, discuss their meaning and how they will be used, and then be diligent about applying them equally to every applicant.
  • Allow sufficient time to evaluate each candidate so reliance on snap judgments and stereotypes has less influence.
  • Seek advice from individuals who are different from you when evaluating candidates.
  • Always have at least two individuals separately evaluate each candidate, and consider using an agreed-upon rating scale to independently weigh each selection criteria.