6.3: Avoiding Bias in Selection
- Page ID
- 47057
\( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)
\( \newcommand{\id}{\mathrm{id}}\) \( \newcommand{\Span}{\mathrm{span}}\)
( \newcommand{\kernel}{\mathrm{null}\,}\) \( \newcommand{\range}{\mathrm{range}\,}\)
\( \newcommand{\RealPart}{\mathrm{Re}}\) \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)
\( \newcommand{\Argument}{\mathrm{Arg}}\) \( \newcommand{\norm}[1]{\| #1 \|}\)
\( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)
\( \newcommand{\Span}{\mathrm{span}}\)
\( \newcommand{\id}{\mathrm{id}}\)
\( \newcommand{\Span}{\mathrm{span}}\)
\( \newcommand{\kernel}{\mathrm{null}\,}\)
\( \newcommand{\range}{\mathrm{range}\,}\)
\( \newcommand{\RealPart}{\mathrm{Re}}\)
\( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)
\( \newcommand{\Argument}{\mathrm{Arg}}\)
\( \newcommand{\norm}[1]{\| #1 \|}\)
\( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)
\( \newcommand{\Span}{\mathrm{span}}\) \( \newcommand{\AA}{\unicode[.8,0]{x212B}}\)
\( \newcommand{\vectorA}[1]{\vec{#1}} % arrow\)
\( \newcommand{\vectorAt}[1]{\vec{\text{#1}}} % arrow\)
\( \newcommand{\vectorB}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vectorC}[1]{\textbf{#1}} \)
\( \newcommand{\vectorD}[1]{\overrightarrow{#1}} \)
\( \newcommand{\vectorDt}[1]{\overrightarrow{\text{#1}}} \)
\( \newcommand{\vectE}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash{\mathbf {#1}}}} \)
\( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)
\(\newcommand{\avec}{\mathbf a}\) \(\newcommand{\bvec}{\mathbf b}\) \(\newcommand{\cvec}{\mathbf c}\) \(\newcommand{\dvec}{\mathbf d}\) \(\newcommand{\dtil}{\widetilde{\mathbf d}}\) \(\newcommand{\evec}{\mathbf e}\) \(\newcommand{\fvec}{\mathbf f}\) \(\newcommand{\nvec}{\mathbf n}\) \(\newcommand{\pvec}{\mathbf p}\) \(\newcommand{\qvec}{\mathbf q}\) \(\newcommand{\svec}{\mathbf s}\) \(\newcommand{\tvec}{\mathbf t}\) \(\newcommand{\uvec}{\mathbf u}\) \(\newcommand{\vvec}{\mathbf v}\) \(\newcommand{\wvec}{\mathbf w}\) \(\newcommand{\xvec}{\mathbf x}\) \(\newcommand{\yvec}{\mathbf y}\) \(\newcommand{\zvec}{\mathbf z}\) \(\newcommand{\rvec}{\mathbf r}\) \(\newcommand{\mvec}{\mathbf m}\) \(\newcommand{\zerovec}{\mathbf 0}\) \(\newcommand{\onevec}{\mathbf 1}\) \(\newcommand{\real}{\mathbb R}\) \(\newcommand{\twovec}[2]{\left[\begin{array}{r}#1 \\ #2 \end{array}\right]}\) \(\newcommand{\ctwovec}[2]{\left[\begin{array}{c}#1 \\ #2 \end{array}\right]}\) \(\newcommand{\threevec}[3]{\left[\begin{array}{r}#1 \\ #2 \\ #3 \end{array}\right]}\) \(\newcommand{\cthreevec}[3]{\left[\begin{array}{c}#1 \\ #2 \\ #3 \end{array}\right]}\) \(\newcommand{\fourvec}[4]{\left[\begin{array}{r}#1 \\ #2 \\ #3 \\ #4 \end{array}\right]}\) \(\newcommand{\cfourvec}[4]{\left[\begin{array}{c}#1 \\ #2 \\ #3 \\ #4 \end{array}\right]}\) \(\newcommand{\fivevec}[5]{\left[\begin{array}{r}#1 \\ #2 \\ #3 \\ #4 \\ #5 \\ \end{array}\right]}\) \(\newcommand{\cfivevec}[5]{\left[\begin{array}{c}#1 \\ #2 \\ #3 \\ #4 \\ #5 \\ \end{array}\right]}\) \(\newcommand{\mattwo}[4]{\left[\begin{array}{rr}#1 \amp #2 \\ #3 \amp #4 \\ \end{array}\right]}\) \(\newcommand{\laspan}[1]{\text{Span}\{#1\}}\) \(\newcommand{\bcal}{\cal B}\) \(\newcommand{\ccal}{\cal C}\) \(\newcommand{\scal}{\cal S}\) \(\newcommand{\wcal}{\cal W}\) \(\newcommand{\ecal}{\cal E}\) \(\newcommand{\coords}[2]{\left\{#1\right\}_{#2}}\) \(\newcommand{\gray}[1]{\color{gray}{#1}}\) \(\newcommand{\lgray}[1]{\color{lightgray}{#1}}\) \(\newcommand{\rank}{\operatorname{rank}}\) \(\newcommand{\row}{\text{Row}}\) \(\newcommand{\col}{\text{Col}}\) \(\renewcommand{\row}{\text{Row}}\) \(\newcommand{\nul}{\text{Nul}}\) \(\newcommand{\var}{\text{Var}}\) \(\newcommand{\corr}{\text{corr}}\) \(\newcommand{\len}[1]{\left|#1\right|}\) \(\newcommand{\bbar}{\overline{\bvec}}\) \(\newcommand{\bhat}{\widehat{\bvec}}\) \(\newcommand{\bperp}{\bvec^\perp}\) \(\newcommand{\xhat}{\widehat{\xvec}}\) \(\newcommand{\vhat}{\widehat{\vvec}}\) \(\newcommand{\uhat}{\widehat{\uvec}}\) \(\newcommand{\what}{\widehat{\wvec}}\) \(\newcommand{\Sighat}{\widehat{\Sigma}}\) \(\newcommand{\lt}{<}\) \(\newcommand{\gt}{>}\) \(\newcommand{\amp}{&}\) \(\definecolor{fillinmathshade}{gray}{0.9}\)- Discuss how to avoid discrimination in the selection process
- Discuss how to avoid perceptual errors (unconscious bias) in the selection process
Avoiding Discrimination
The go-to reference for avoiding discrimination in the selection process is the EEOC’s Uniform Guidelines on Employee Selection Procedures. In brief, the purpose of the Guidelines is “to aid in the achievement of our nation’s goal of equal employment opportunity without discrimination on the grounds of race, color, sex, religion or national origin.”[1] The Guidelines constitute a uniform set of principles governing employee selection procedures that are consistent with applicable legal standards and validation standards generally accepted by the psychological profession.
The Guidelines apply to
- most private and public employers, including labor organizations, employment agencies, state and local governments, and federal government contractors and subcontractors
- all selection procedures used to make employment decisions, including interviews, review of experience or education from application forms, work samples, physical requirements, and evaluations of performance
- employee selection procedures used in making retention, promotion, transfer, demotion, or dismissal decisions
As an aside: the guidelines do not have bearing on recruiting procedures or practices—for example, practices designed to attract members of a particular race, sex, or ethnic group that are under-represented.
The California State Personnel Board’s “Summary of the Uniform Guidelines on Employee Selection Procedures” is more accessible than trying to parse the original Guidelines.[2] To excerpt: The Uniform Guidelines “establish uniform standards for employers for the use of selection procedures and to address adverse impact, validation, and record-keeping requirements…and outline the requirements necessary for employers to legally defend employment decisions based upon overall selection processes and specific selection procedures.”[3]
- Employer policies or practices which have an adverse impact on employment opportunities of any race, sex, or ethnic group are said to be discriminatory and are illegal unless justified by business necessity.
- If adverse impact exists for a specific selection procedure or an overall selection process, the selection procedure must be validated to document the relationship between the procedure and successful performance on the job.
- Discrimination or adverse impact is decided based on the “4/5ths” or “80 percent” rule. If a comparison group does not have a passing rate equal to or greater than 80 percent of the passing rate of the highest group, then it is generally held as evidence of adverse impact for the particular selection procedure.
- If the use of a particular selection procedure results in adverse impact, the employer can cease use of the procedure or it must demonstrate the “business necessity” of the selection procedure through validation.
- Validation is the establishment of a clear relationship between a selection procedure and the requirements of successful job performance. Refer to the Summary for additional details on the aspects of validity testing and additional requirements.
For a higher-level view, the following is a representative sample of the EEOC’s Employer Best Practices for Testing and Selection:[4]
- Employers should administer tests and other selection procedures without regard to race, color, national origin, sex, religion, age (40 or older), or disability.
- Employers should ensure that employment tests and other selection procedures are properly validated for the positions and purposes for which they are used. The test or selection procedure must be job-related and its results appropriate for the employer’s purpose. While a test vendor’s documentation supporting the validity of a test may be helpful, the employer is still responsible for ensuring that its tests are valid under UGESP.
- If a selection procedure screens out a protected group, the employer should determine whether there is an equally effective alternative selection procedure that has less adverse impact and, if so, adopt the alternative procedure.
- To ensure that a test or selection procedure remains predictive of success in a job, employers should keep abreast of changes in job requirements and should update the test specifications or selection procedures accordingly.
- No test or selection procedure should be implemented “casually”; that is, without an understanding of its effectiveness and limitations for the organization, its appropriateness for a specific job, and whether it can be appropriately administered and scored.
Avoiding Perceptual Errors
When reviewing a final slate of candidates, it’s important to be aware of the potential for perception errors and unconscious bias.
Research has indicated that interviewers make decisions about candidates rapidly—within the first 30 seconds to 2 ½ minutes, to be precise.[5] Unfortunately, we also tend to overrate our ability to evaluate others. In a study cited in Fundamentals of Human Resource Management, an algorithm that weighed several objective job-related criteria did a 25% better job of selecting successful candidates than experienced managers. A common mistake is judging candidates based on a first impression or “likeability.” As IBM Smarter Workforce business development executive Jason Berkowitz notes: “It’s so easy to assume that a firm handshake and good eye contact means someone is competent across the board.”[6] Interview tip: Don’t use the interview to try to validate your initial judgment—positive or negative. Berkowitz’s recommendation: “Hiring managers should actually try to disprove their initial impression.”[7]
Another common error is interviewer bias, where the interviewer allows information reviewed prior to the interview—resume, test scores, social media activity—to shape their perception of the candidate. In order to counter this and related judgement errors, interviewers should cultivate active listening skills and focus on evaluating each candidate relative to the same standards.
Another source of evaluation error is unconscious bias, which is “a prejudice in favor of or against one thing, person, or group compared with another—usually in a way that’s considered to be unfair.”[8] In a Fast Company article titled “How Unconscious Bias Affects Everything You Do,” author Howard Ross notes that “Over 1,000 studies in the past 10 years alone have conclusively shown that if you’re human, you have bias, and that it impacts almost every variation of human identity: Race, gender, sexual orientation, body size, religion, accent, height, hand dominance, etc.”[9] His conclusion: “The question is not ‘do we have bias?’ but rather ‘which are ours?’” The implications for human resource management? How do we design selection processes that counter unconscious bias?
The classic example of how to design around unconscious bias is that of major orchestras attempting to overcome systemic hiring bias and achieve relative gender equity. Although there were a number of factors that contributed to bias—for example, issuing private invitations rather than advertising auditions—the critical modification was implementing blind auditions, where raters did not see the musicians.
Although this is where most people end the story, the curtain wasn’t the final bias hack. As University of Tromsøt professor Curt Rice reported in the Guardian, the curtain did make it 50% more likely that a woman would advance to the finals; however, the sound of a woman’s heels on the stage still triggered unconscious bias. The final modification was to have musicians remove their shoes before entering the audition area.[10] Rice observes that while gender blind evaluations may seem impractical, orchestras prove it can be done. His challenge: “If we trust the research and accept that women are being judged more fairly because of the screen, perhaps we should ask if there’s any way to replicate the musicians’ success. What kind of screen would be needed at your workplace?”
Unconscious bias can also be a factor in the pre-screening process. In an article titled “Can dogs help us avoid hiring bias?”[11] BBC Senior Editor Meredith Turits describes TED director of research and development Aaron Weyenberg’s attempts to hack unconscious bias in the hiring process. One of the hiring policies he implemented was to conduct first-round interviews with audio-only telephone calls, to eliminate consideration of irrelevant information. However, he was aware that LinkedIn “often finds its way into the process before that. And what that does is expose to me to information I actually don’t want and doesn’t help me – like their appearance (and thereby their approximate age), name, et cetera.” To get around that, Weyenberg used the Profile of Dogs Chrome browser extension, which replaces a LinkedIn user’s profile picture with randomly assigned image of a dog’s face. The problem with that theory, as Princeton assistant chair of psychology Alexander Todorov explains, is that people make associations with all kinds of things. Even dogs. For instance, different dogs have different reputations and invoke different responses (think: pit bull versus golden retriever) and the viewer may transfer those associations to the candidate—even when they know the image is randomly generated. Turits notes that even if we are able to remove bias from an initial screening, it’s likely to slip back in at a later stage. As Todorov phrases it: “the mind is a big associative machine.” Inclusion consulting firm Jones Diversity CEO Sharon Jones notes that “Most people believe they are fair and support a meritocracy in hiring and in the workplace. What people don’t understand that wanting to be fair doesn’t make you fair.”[12]
So if the goal is to neutralize unconscious bias and believing we’re fair doesn’t make it so, what does? Ross proposes two approaches: awareness and design. Specifically, he notes that “when we are aware of our biases and watch out for them, they are less likely to blindly dictate our decisions.”[13] To the second point, he advocates for developing approaches that help us make decisions more consciously. One of the specific actions he recommends is priming, explaining that “by consciously priming people to pay attention to potential areas of bias, we have found that they can be encouraged to be more conscious of their decision-making processes.” Specifically, the goal is to help managers identify what they’re reacting to and refocus on information that’s relevant to the job.
- "Uniform Guidelines on Employee Selection Procedures." Accessed July 26, 2019. ↵
- "Summary of the Uniform Guidelines on Employee Selection Process." California State Personnel Board. Accessed July 26, 2019. ↵
- Ibid. ↵
- "Employment Tests and Selection Procedures." The U.S. Equal Employment Opportunity Commission. Accessed July 26, 2019. ↵
- "What is First Impression Error?" The Headhunters. Accessed July 26, 2019. ↵
- Trinder, Elizabeth. "5 Common Hiring Mistakes and How to Avoid Them." Agent Entrepreneur. June 11, 2014. Accessed July 26, 2019. ↵
- Ibid. ↵
- "Unconscious Bias." UCSF Office of Diversity and Outreach. Accessed July 26, 2019. ↵
- "How Unconscious Bias Affects Everything You Do." Fast Company. October 22, 2014. Accessed July 26, 2019. ↵
- Rice, Curt. "How Blind Auditions Help Orchestras to Eliminate Gender Bias." The Guardian. Accessed July 26, 2019. ↵
- Turits, Meredith."Can Dogs Help us Avoid Hiring Bias?" BBC Worklife. February 10, 2019. Accessed July 26, 2019. ↵
- Ibid. ↵
- "How Unconscious Bias Affects Everything You Do." Fast Company. October 22, 2014. Accessed July 26, 2019. ↵