Eliminating bias in selection

Eliminating bias in selection

by Felicity Menzies

Selection procedures refer to any measure or procedure or a combination of measures or procedures used as a basis for an employment decision. Selection procedures include a range of assessment techniques from traditional paper and pencil tests, work samples and other skills-based tests, psychometric tests such as personality or aptitude tests, probationary periods, educational and work experience requirements, informal or casual interviews, structured interviews, and interview scorecards or skills rubrics.

Selection methods vary with respect to their propensity for bias to influence candidate assessment. Although the most common selection procedure, a vast body of research shows that informal unstructured interviewing is the least objective selection method. Cognitive neuroscience research has taught us that most decisions we make, especially regarding people, are “alarmingly contaminated” by biases that operate under the scope of human consciousness. The existence of implicit or hidden biases means that we often make decisions and assessments of others that we believe are consistent with our conscious intentions to be fair and objective, but in fact, our unconscious beliefs and preferences are driving our responses. Unconscious bias in selection has profound implications— when assessing a candidates suitability for a role, we may be adding our own subliminal and emotional criteria to that decision. Criteria we might not even be aware of and which may have no basis in facts. Because of the propensity for bias to influence candidate assessment in unstructured interviews, in the majority of cases, candidate interviews add zero credibility to the hiring decision. In the best cases, unstructured interviews add only 10% validity to the process. Despite these findings, informal interviews remain a preferred selection procedure, largely because recruiters commonly overestimate their interviewing skills and the predictive effectiveness of their candidate assessments.

Bias can also be institutionalised in the form of selection processes and practices that systematically favour some groups over others and are part of the fabric of an organisation. Examples include the practice of ‘tap-on-the shoulder’ appointments that allow for affinity bias in selection decisions.

Costs of Bias in Selection

Good hiring is critical for business performance. Selecting the best candidates not only drives higher performance—but it also reduces costs. Harvard Business Review report that 80% of turnover is due to bad hiring decisions. Research from the Society for Human Resource Management concludes that a poor hire can cost a company up to five times the amount of the hire’s salary when one considers the original cost of hiring the employee, the cost of lost productivity, the cost of training the poor hire, and the cost of hiring and training a replacement, among other factors.

If you have biases which are impacting recruitment decisions you are likely not choosing the best person for the job. In addition, you are likely to end up with a relatively homogenous workplace, missing out on the competitive potential of a diverse workforce.

Strategies for Eliminating Bias From Selection

Although 81% of employers recognise the potential for unconscious bias to influence selection, 42% of employers do not use any strategies to reduce bias in selection.

Eliminating bias from selection involves techniques and practices that;

(i)      encourage the assessor to focus on matching demonstrable applicant capabilities to well-defined role requirements, and

(ii)     limit the potential for information about the candidate that is not relevant for job performance but that might influence how the assessor personally feels about or perceives the candidate to impact the assessor’s judgment regarding candidate-role fit.

Examples of effective strategies for minimising bias in selection are detailed below. Employers committed to fair selection can choose to employ one, many, or all of suggested techniques to improve the objectivity of their selection decisions. When combined with inclusive recruitment, employers that employ multiple methods of eliminating bias from selection stand the best chance of hiring the best candidates and building a diverse workforce.

  •      Well-defined job descriptions with objective, specific, and relevant selection criteria
    •      identify the specific tasks and responsibilities required for the role. Rank those tasks in terms of how frequently they are carried out by the employee and how critical they are to succeed in the role. Clarity on what tasks are most critical for performance and are performed most often helps assessors to map candidates strengths to the tasks that are most important for success in the role.  
    •      specify selection criteria (e.g. knowledge, skills, traits, qualifications, experience) that map to the role requirements, not to the type of person who you think would be successful in that role. For example, does the candidate necessarily need to be university educated, outgoing, or ambitious?
    •      rank selection criteria as ‘non-negotiable’, ‘desired’, ‘can be developed’, ‘can be accommodated for’. Distinguishing between criteria that are essential and those that are non-essential or can be developed on-the-job encourages consideration of a broader pool of candidates.
  •      Structured application forms
    •      design candidate application forms that specifically request information pertaining to objective selection criteria. This avoids the risk of the assessor being influenced by information provided in a CV that might not be relevant to the role.
  •      Blind and automated shortlisting
    •      when shortlisting candidates, remove all identifying information from application forms or CV’s that might unconsciously bias assessment of the candidate. For example, name (can indicate gender or ethnicity), gender, age, educational background, interests, volunteer work, place of residence. While blind shortlisting can be laborious when performed manually, software solutions can make removing identifying information from CV’s efficient and feasible. Blind shortlisting is only useful for improving diversity hiring, however, if the candidate pool is diverse. Blind shortlisting of a homogenous applicant pool will not lead to more diverse hires.
    •      technology can also be used to efficiently shortlist candidates by matching job criteria to detail provided by the candidate in the application form
  •      Phone screening of applicants
    •      similar to blind CV’s, candidate screening by phone helps to minimise the influence of bias by decreasing the visibility of some diversity dimensions that are not relevant for job performance but might influence the assessor’s evaluation of a candidate in an unconscious manner. Phone screens are a way of making interviews ‘blind’ and encourage the assessor to focus in the first instance on skills, abilities, and knowledge relevant to the role rather than other characteristics of the candidate that can unconsciously influence assessment. Technology can be helpful in managing phone screening. Interview.io, for example, allows candidates and companies to conduct completely anonymous interviews. If the company is interested, the candidate is notified. They can then “unmask” themselves to reveal their identity. 
    •      while phone interviews minimise the influence of visual first impressions, they do not eliminate accent or linguistic bias. If you want to make the interview even more blind, you can go so far as to use a voice modulation app for technical interviews.
  •      Inclusive selection processes
    •      be transparent regarding the interview or other assessment processes to allow the candidate to prepare. Useful information includes the format of the interview or other assessment, the number of stages of the hiring process, timelines for response and feedback, who they will be meeting with, directions on how to get to the venue, and suggestions on how to prepare for the interview or assessment including a copy of questions they will be asked. Providing transparency is particularly important for diverse talent who may experience increased levels of anxiety regarding the assessment process. 
    •       check for preferred pronoun use and use the name the candidate gives in the application form
    •      don’t start the interview right away. Walking around the office or to a café will help to put the candidate at ease and improve interview performance.
    •      always conduct interviews in a warm and friendly tone and convey respect to all candidates. Even if that person doesn’t get an offer, they may tell a friend or two if they get a really positive or really negative impression.
    •      establish and inform candidates of grievance procedures regarding recruitment, assessment, and selection
    •      traditional assessment methods such as on-site, face-to-face interviews or work samples can disadvantage candidates with a disability, culturally and linguistically diverse candidates, introverts, candidates with non-traditional work schedules, and geographically dispersed candidates, among others. Employers can help to level the playing field for diverse candidate pools by asking if the candidate needs any reasonable adjustments at interview, for example, accessibility to the building, information, materials and communication; interpreter services; flexibility in scheduling; allowing extra time for to prepare for an interview or complete an assessment; or pre-recorded or live video interviews. 
  •      Pre-recorded interviews
    •      pre-recorded interviews make the selection process more flexible and address some of the barriers facing individuals with non-traditional work schedules, candidates that are geographically dispersed, candidates with mobility restrictions or other disabilities, and candidates who are unable to take time off from their existing employment to attend interviews. Candidates have the opportunity to record themselves at home, on their own time. The employer creates a list of interview questions – either via text or video – and candidates record themselves responding to these questions. Interviewers can then watch these pre-recorded interviews on their own time, which also removes scheduling barriers. If multiple interviewers are involved in the process, each interviewer can watch the candidate’s pre-recorded interview on their own schedule. By incorporating video interviewing into its graduate recruitment, Goldman Sachs has extended its reach beyond Ivy League schools and enabled the firm to interview a much larger and diverse pool of candidates. Knowing that an Ivy League accreditation was not essential for success at Goldman Sachs, video interviewing enabled the firm to shift its focus from interviewing a limited number of graduates from Ivy League schools face-to-face to interviewing a larger number of top graduates irrespective of where they went to university. 
  •      Behavioural-based interviewing
    •      behavioural-based interviewing allows candidates an opportunity to demonstrate their competencies (knowledge, skills, and abilities) by describing how they achieved examples of success in their current or past roles, how they managed challenges, or by explaining how they would approach a hypothetical situation. The principle behind behaviour-based interviewing is that past behaviour is an effective indicator of future behaviour. Interviewers should develop behaviour-based questions that tap the traits and skills they consider necessary for succeeding in the role. This can be explored by asking questions such as, “Describe a difficult work-related problem that required you to come up with a creative solution. Tell me the steps you took and why”, or “Please give a specific example of when you collaborated with another individual. How would you evaluate or describe the results from that effort?” Behavioural-based interviewing will help you avoid making selections based on assumptions or intuitions or solely on credentials. As part of a recruitment overhaul, Goldman Sachs designed structured interview questions to assess candidates on 10 core competencies, including analytical thinking and integrity, which were known correlate with long-term success at the firm. Candidates are evaluated on six competencies in the first round of video interviews; if they progress, they’re assessed on the remaining four during in-person interviews.
  •      Interview scorecards  
    •       the predictive validity of interviews can be improved by employing a rating system for scoring candidate answers to the questions asked by the interviewer. To reduce inconsistency in ratings across different interviewers and candidates, the rating system should describe in as much detail as possible how to score a candidates answer.
    •      when interview scorecards are used, ensure selection decisions are supported by the data collected and recorded on the scorecards. Do not use any casual conversation that may have taken place outside of the structured interview.
  •      Consider diversity as a rating criteria
    •      include diversity as a selection criteria, where relevant. For example, recognise the value of cultural and linguistic skills for customer-facing roles.
  •       Take a whole-of-person approach 
    •      do not rely solely on experience, credentials, and career progression. Consider a candidate’s journey and what they have learned from their experiences. Assessing competence on the basis of quick successive promotions might lead you to overlook the value of grit and perseverance, for example. If a person took a risk and it did not pay off, for example, they may have learned more from that challenge than a person who took a safer path. The lessons people learn throughout their careers are often undervalued in the selection process but can contribute significantly to future performance. 
  •       Take an additive approach
    •       identify the mindsets, skills, and diverse experiences that drive execution excellence for your team seek individuals who bring those skills or can fill any skill gaps in your team. Ask  ‘How can [or does] this person add to the total value (composition) of our team?”, “What skills and experiences am I missing on my team that this person has?” Be careful not to focus on one-dimensional characteristics. For example, don’t determine you need a woman balance out the team. Diversity for diversity’s sake often leads others to make negative assumptions about your people decisions — and about those you hire or promote. Instead, look at how people can add to the total portfolio of mindsets, skills, and experiences on the team.
  •      Rate each candidate independently of others
    •      interviewers should not look back at how they rated other candidates until all of the interviews have been completed to ensure that a rating for one applicant doesn’t affect the rating of another
  •      Take notes during the interview
    •      the interviewer should score each answer immediately after it is provided. Studies show a recall bias for vivid examples such as stories and also for answers that are most recent. Waiting until the end of the interview to rate answers risk forgetting an early or less-vivid but high-quality answer, or favouring candidates whose speaking style involves storytelling or whose latter responses are particularly strong even if their earlier responses are weaker than other candidates.
    •      studies also show a universal tendency for us to pay greater attention to and recall information that confirms our preexisting ideas and to dismiss information that is inconsistent with our preexisting ideas and prejudgements. Researchers label this tendency confirmation bias. Taking notes helps to counter our tendency to pay greater attention to and recall information congruent with our biases. 
  •       Unconscious bias training for hiring managers
    •      unconscious bias is universal—we all have bias in some shape or form. It is an inevitable consequence of cognitive processes and social conditioning. However, although bias is evitable, acting on it isn’t. Research has shown that we can override our reflexive responses with controlled and conscious thought or reflection. When we are motivated to be fair and unprejudiced because of either a strong internalised belief that it is morally correct to treat others fairly or because of strong social norms and legal restrictions against expressed prejudice and discrimination, we can engage controlled mental processes to override biased automatic responses. Unconscious bias training raises awareness of bias and transfers strategies for hiring managers for managing their biases. In a pilot project led by the Victorian State Government, researchers found that unconscious bias training resulted in a statistically significant increase in employee-reported self-efficacy and positive intentions regarding diversity hiring.
  •      Work samples and other skills-based testing    
    •      give a work sample test that mimics the kinds of tasks the candidate will be performing on the job. Goldman Sachs, for example, offers a technical coding and math exam for engineering applicants.  For roles that do not require skills needed for strong interview performance, work samples help to level the playing field for individuals who do not perform well in interviews or who have little prior experience by allowing them an opportunity to showcase the skills that directly relate to the job tasks. Research shows that work samples are the best indicators of future job performance and outperform interviews, unstructured and structured. 
    •      whenever possible, ask for samples or examples of work, and review them before you meet with the candidate. This will help you evaluate the candidate’s actual skill first, and their emotional impression on you next, helping the diffuse the impact of biased first impressions.
  •      Validated assessment tools
    •      use validated tests for predicting job performance such as emotional intelligence, cultural intelligence (for performance in diverse cultural settings), general cognitive ability (IQ), personality assessments and other psychometric tools. Using validated assessment tests controls for unconscious bias and improves objectivity in candidate assessment, which in turn drives greater diversity. A study of 150 companies found that those that used a personality assessment in their hiring had more racially diverse workforces.
  •      Panels of independent and diverse assessors
    •       by involving others who ideally represented the diversity that the employer is striving to achieve, recruiters are exposed to different perceptions and opinions about the candidate and are challenged to justify their assessment of a candidate with objective data. This helps to decrease the potential for emotion-driven or instinctive decision-making.
    •      panel members should always score candidates individually first and save their discussion until the end of the interview process. Independent ratings guard against conformity bias (groupthink). Conformity bias is a tendency to be influenced by ideas and behaviours of others rather than exercising our own independent judgment.  
    •      reference checks are another data point for the hiring process but it is important to remember that referee assessments can be biased. Asking referees for examples of achievements on the job that demonstrate behaviours required in the role as well as probing for any red-flags can improve the objectivity of referee data. Ideally, the reference checks should be conducted by an individual who did not interview the candidate. This avoids the person conducting the reference check being influenced by confirmation bias.
  •      Multiple assessment methods
    •      combine structured interviews with work samples, validated psychometric tools, panels, and reference checks. Using multiple assessment methods offers multiple data points and decreases the risk of bias. 
  •      After the interview
    •      help to secure acceptance of your offer of employment by giving candidates an opportunity to chat with current members of the workforce. This is particularly important for women and minorities who may be uncertain of how welcoming and supportive the wider organisation is to difference. Introduce diverse candidates to diversity champions, role models or representatives from Employee Affinity Groups.
    •      for candidates that are not successful, always give objective interview feedback. This helps to minimise any concerns that the candidate might have regarding the role of bias in selection.
    •      undertake an evaluation process of successful candidates to determine if their expectations of the role and department were aligned with their actual experiences after joining the organisation
    •      survey all shortlisted candidates to understand individual perceptions of the recruitment process
  •      Metrics
    •      diversity metrics should be established and tracked for the purposes of identifying bias risk areas and opportunities. Example metrics include candidates shortlisted, candidates performance at interview, candidate selection, candidate retention. For example, if a hiring manager gives 10 interviews, five to men and five to women, and four out of the highest five scores are given to males, there should be an investigation into whether there might be a pro-male bias in the interview process. The investigation might find the pattern of scores is supported, but it might uncover a bias that the employer can address. Vodafone ran a pilot in India to test the effect of removing gender from the CVs of job applicants. Local managers had assumed they were failing to appoint women into tech roles because of a small pool of female candidates. The data, however, told another story—a significant number of highly qualified women were applying but were not getting interviews. Similarly, at professional services firm KPMG, crunching the numbers on internal promotions revealed that proportionately more men than women were being promoted to senior roles. This was not simply a matter of male bosses appointing men in their own image. The firm found that whereas the men would apply for a role if they had 80 percent of the [required] skills, women would think they were missing 20 percent and not bother. The firm responded by requiring that, when a promotion is advertised, line managers are encouraged to check whether their high-potential female colleagues have applied and if not, ask why.
    •      metrics not only help an organisation identify bias in the recruitment and selection process, but they also offer employers a means of tracking the success of any initiatives to improve the objectivity of hiring processes, turning recruiting departments into a laboratory for continuous learning and refinement. All assessment methods, including the questions asked in structured interviews, should be tracked for their predictive success. Assessment methods or questions that are most positively correlated with future job performance should receive the highest weighting in hiring decisions. If the analysis shows that a method or question lacks predictive validity, it should be deleted from the assessment process and the employer should experiment with alternative measures.
    •      when assessing selection methods for predictive validity, however, caution is advised. There is a risk that bias across the employee life-cycle may corrupt measures of predictive validity. Say, for example, only White men are promoted to leadership roles, then assessment measures that favour White men will rank more highly on predictive validity. Scientists have found evidence that data-driven recruitment algorithms have the potential to learn our prejudices. Unless carefully monitored, their filters detect patterns of underrepresentation and reproduce them, perpetuating biases against already disadvantaged groups such as older and disabled workers, women and ethnic minorities.
    •      diversity metrics can also be used to drive accountability. For example, diversity metrics for candidate pools or shortlists forces recruiters to look beyond their biases are to seek a high-quality diverse candidate pool.

Felicity Menzies is CEO and Principal Consultant at Include-Empower.Com, a diversity and inclusion consultancy with expertise in inclusive leadership, unconscious bias, cultural intelligence and inclusion, gender equity, empowering diverse talent. Felicity is an accredited facilitator with the Cultural Intelligence Centre and the author of A World of Difference. Felicity has over 15 years of experience working with and managing diverse workforces in blue chip companies and is a Fellow of Chartered Accountants of Australia and New Zealand. Felicity also holds a Bachelor of Commerce and a Bachelor of Arts in Psychology.