The job interview is a peculiar institution. It asks two parties who have never met to assess each other's suitability for a relationship that will define a significant portion of each party's waking life, typically over a period of thirty to ninety minutes, using a highly stylized conversational format that both parties know is not representative of how either of them normally behaves.

The research on whether job interviews predict job performance is, to put it charitably, mixed. Traditional unstructured interviews — where the interviewer decides what to ask and how to interpret the answers — have been shown in meta-analyses to predict job performance barely better than chance. The interviewer's impression is shaped more by factors uncorrelated with competence — physical appearance, verbal fluency, cultural familiarity, the interviewer's own cognitive biases — than by accurate assessment of the candidate's ability to do the job.

Structured interviews, where all candidates are asked the same questions and answers are evaluated against predetermined criteria, perform significantly better. Behavioral interviews — where candidates are asked to describe specific past situations that demonstrate relevant skills — perform better still. The irony is that the formats with the best predictive validity are often the least comfortable for both parties: they feel less like conversation and more like examination.

From the candidate's perspective, the interview is a performance with a specific audience. The audience wants to feel confident about this candidate — confident that she can do the work, that she will fit into the team, that she will not create problems that will reflect badly on the interviewer's judgment. The candidate's job is not to present an accurate picture of herself but to manage the interviewer's confidence effectively.

This is not dishonesty. It is the social skill that the interview format rewards, and it is a legitimate skill for a job candidate to develop.