Job Interviews Don’t Work

Farnam Street, 2020-07-11

Problem Statements

  • Job interviews are not best predictors of performance (more on that later).

  • Maximizing the role of bias and minimizing the role of evaluating competencies.

  • Encourage homogenous environment and groupthink.

  • If a role doesn’t require charm (salespeople), unstructured interviews (i.e. free-form without agenda) are biased towards charismatic people (many of whom are liars).

  • Perceived cultural fit is good for small firms, but not for large companies.

Ineffective Interviews

Bias (age, appearance, gender, etc.) is very much present – interviewers look for people like them.

o   Attractive people look more qualified;

o   Tall people (esp. men) are considered better leaders.

o   Deep voice == trustworthy.

  • Bias —> Decision to hire —> backward rationalization [confirmation bias].

  • Answers to interview questions are poor predictors of real work. Fundamental attribution error: people’s behaviours are thought to be a correct reflection of their selves, ignoring the impact of circumstances.

  • Bending the truth is common:

o   Embellishment – take credit for things we haven’t done

o   Tailoring – adapt answers to fit the job requirements

o   Constructing – piece together different experiences to provide better answers.

  • Intuition works well for areas with fast feedback and clear cause and effect.

o   Job feedback take a long time. The person “at fault” is usually the employee, not the interviewer.

  • Interviews generate lots of noise —> less relevant information, but more overconfidence.

o   Discard information contradicting our intuition

o   Less challenging information to process feels better (but not correct). Don’t make me think!

  • Expertise is about having a predictive model of something, not just knowing a lot about it. Not always possible to get that during the interview.

Making Interviews More Effective

  • Structured Interviews. Everyone gets the same questions with the same wording, no improvisation.

o   Consistent scoring model reduces [not eliminates] bias. But substantially reduces noise.

  • Blind auditions. Disguise key info leading to biased decisions. Focus on skills alone.

  • Competency-related evaluations. Ask the candidate to carry out the tasks that are part of the job (or do sample development tasks in IT firms). Many companies still have it wrong (in-person interviews override competency evaluations).