The last of the ideas from Daniel Kahneman’s book, Thinking, Fast and Slow, I’m going to write up is the superiority of algorithm-driven interview techniques in personnel selection. This is related to his thinking about when to trust intuition I wrote up in a previous post.
Kahneman says that hundreds of studies have been completed comparing the effectiveness of clinical vs. statistical predictions of long-term mental health in psychology. Sixty percent show statistical prediction is better; the remainder show no difference, but statistical prediction is much cheaper.
Kahneman believes, therefore, that for predictions in similarly “noisy” environments (he calls them “low-validity”)—such as personnel decisions—we should also use algorithms rather than intuitions to make our selections. He thinks algorithms are superior for four reasons: