Week 22 Day 1: Most Interviews Measure Confidence, Not Competence

Interviews reward people who are good at interviews. That is a different skill from being good at the job. The candidate who speaks fluently, maintains eye contact, and tells polished stories will consistently outperform the candidate who is quietly brilliant but does not interview well. The result is a systematic bias toward charisma over capability -- and most hiring managers do not realize they are doing it.

I hired a senior engineer once who gave the best interview I had ever seen. Clear, structured answers. Compelling stories about past projects. Strong opinions about architecture. He checked every box on the scorecard. Within three months, it was clear he was a performer, not a doer. His answers in the interview were not lies -- they were carefully curated highlights from a career of marginal contributions to other people's work. He could describe solving problems with extraordinary clarity because he had watched other people solve them and had absorbed the narrative without doing the work. Meanwhile, the candidate I did not hire -- because she gave 'average' interview answers -- went on to become a principal engineer at another company. Her interview answers were average because she was thinking in real time rather than reciting rehearsed stories. I was measuring fluency, not substance. After that experience, I restructured my interviews entirely. I stopped asking questions that could be rehearsed and started asking questions that required real-time problem-solving, self-reflection, and honest uncertainty. The shift did not just change who I hired -- it changed the quality of every team I built afterward.

The confidence-competence confusion in interviews is a well-documented phenomenon in personnel psychology. Research by Barrick, Shaffer, and DeGrassi (2009) found that interview performance correlates more strongly with extraversion (r = 0.29) and self-monitoring ability (r = 0.24) than with actual job performance (r = 0.14-0.20 for unstructured interviews). Huffcutt and Arthur's (1994) meta-analysis of 114 interview studies found that unstructured interviews had a validity coefficient of only 0.20 for predicting job performance, compared to 0.51 for structured behavioral interviews. The 'performer' pattern described in level_2 maps to what Paulhus, Westlake, Calvez, and Harms (2013) call 'overclaiming' -- the tendency of some individuals to claim knowledge or experience they do not possess, which is positively correlated with interview success but negatively correlated with job performance. Research by Levashina and Campion (2007) on 'impression management in the employment interview' found that faking and embellishment are extremely common (over 90% of candidates engage in some form) and that interviewers detect faking at rates barely above chance (54%). The restructured interview approach aligns with what Schmidt and Hunter (1998) identify in their meta-analysis as the most valid selection methods: structured behavioral interviews (validity 0.51), work sample tests (0.54), and cognitive ability tests (0.51).