Despite evidence to the contrary, many human resource professionals and hiring managers continue to discount the reliability and predictability of psychometric and cognitive pre-employment testing.
Just a few days I gave a presentation to 50 HR professionals at a regional SHRM meeting. One of my four best practice recommendations was pre-employment testing. Not unexpectedly, one individual, a Vice-President of HR, raised her hand and asked, “how safe are personality tests?”
That’s a pretty common question I get. I wish more HR people would ask it. Asking the question at least starts a conversation. Not asking it keeps HR and employers living in a world of darkness and naiveté.
Many people still believe personality tests are illegal and that their use exposes an employer to more risk. I’ve written (along with many colleagues) quite a few blog posts and articles aboutwhy employee assessments are legal but unfortunately doubts still linger. HR and many employment law attorneys raise the risk flag (which trumps the voice of a consultant) and squash any attempts to assess candidates and employees alike.
But a new research paper titled Legal Risk in Selection: An analysis of processes and toolspresented at the Society for Industrial Organizational Psychology conference dispels many of the lingering myths associated with using personality and other employee tests.
The research findings reviewed EEOC and OFCCP cases settled both in and out of court between 1998 and 2010. Two key areas were covered: (1) type of selection test and (2) the hiring process.
Based on the findings, personality and other psychometric tests do carry some risk. But in nearly every case, the challenge did not involve the validity or reliability of the test but how the assessment was used. For example, according to Dr. Charles Handler, one of the most respected authorities on employee selection, “Cases that went to trial around selection devices were decided for the plaintiff only 28% of the time, vs. 68% for those related to the selection process, meaning that process issues are more likely to land an employer in hot water.”
Cases related to inconsistent process accounted for the largest percentage of all process related cases and over half of these were settled prior to court. A whopping 91% of all inconsistent process cases were found to be discriminatory.
Some examples of process related cases that were lost include:
-
In Dennis v Columbia Colleton Medical Center (2002), the U.S. Court of Appeals described the hospital’s selection process as “a peculiarly informal process” because their explanations for not hiring the plaintiff were different from the written job description, giving the decision “a flavor of post-hoc rationalizations.”
-
In Dunlap v Tennessee Valley Authority (2008), the court determined the company’s hiring process was discriminatory because they found 70 counts of manipulating test scores and changing interview and test scores in candidate rankings.
-
In Allen v Tobacco Superstore (2007), the company relied on word of mouth to publicize open positions and had no consistent procedures for advancement; employees simply asked a supervisor to be considered. The court found the word-of-mouth hiring and promotion process — which resulted in a company-wide dearth of Black store managers despite operating in communities with large Black populations — was discriminatory.
In the cases involving employee assessment tests (cognitive ability and psychomotor tests), decisions that favored the rejected candidates failed to demonstrate the job relatedness of the test, resulting in the adverse impact challenge.
-
In EEOC vs NationsBank of Tennessee (2001) a cognitive ability test discriminated against Hispanic employees by requiring English proficiency, a competency that was not required on the job.
-
In EEOC vs. American Airlines (2002), the company used a pre-employment test for meter readers and janitors that had adverse impact against females and measured skills that were not required on the job.
Interviews were also the subject of discrimination claims. And like the hiring process itself, the unstructured, informal traditional poses a much higher risk to the employer than does psychometric pre-employment testing.
Of all the cases that went to trial involving interviews, 50% of the unstructured interviews were ruled discriminatory. Structured interviews create consistency across interviewers and provide a rationale – job relatedness – behind the interview scores that contribute to hiring decisions.
So here’s a checklist of things HR and hiring managers must do to lower the risk of discrimination and improve the success of their hiring process.
1. Use a structured interview. According to Dr. Handler, “if you are not using a structured interview process, you have a problem.” I couldn’t agree more. The casual, off-the-cuff interview is not only poor risk management but not very predictable when selecting employees.
2. Be consistent. Interview questions must be consistent with job relatedness. Even if the interview questions are structured and managers trained in behavioral interviewing, it doesn’t mean the questions are job related (Dennis v Columbia Colleton Medical Center (2002). The same goes for the job board ad or word of mouth referral programs you use. If you write an ad or ask a question related to a responsibility or skill that is not required for the job, you open the door wider for adverse impact claims to step in.
3. Remain objective. Pre-employment testing IS legal. It’s also a best practice with positive results reported time and time again. Testing is not an astrology or voodoo-like experience but a scientifically proven practice that leads to better hiring results without increasing the risk of adverse discrimination.
Using pre-employment tests for the right reasons (job-relatedness) is the equivalent of having a skilled, unbiased, third party manager interview candidates. Many pre-employment assessments also include job-related interview questions, based on candidate results. These questions structure the interview and keep the focus on job-relatedness.