Today I feel scientifically vindicated. I have often said that bad candidates over-rate their abilities, while good ones under-rate their abilities. If you ask two job applicants "On a scale from 1 to 10, how proficient are you with Excel?" the more skilled person will often rate themselves lower than the less skilled person. I know this because I used to work in a place that tested people's computer skills after they rated themselves.
But today, Jason Seiden wrote a post about the Dunning-Kruger effect - where Cornell scientists showed that Darwin was right when he said, "ignorance more frequently begets confidence than does knowledge." They went on to note that and "people with true knowledge tend to underestimate their competence." Sweet, sweet scientific vindication for one of my long-held opinions.
But if less competent people are often more confident, and are more likely to over-rate their abilities - then asking people to rate themselves is a virtually foolproof way to hire the wrong person. Frankly, if you ask people to rate themselves on their abilities, it would surprise me if you ever hire anyone competent to do the job.
So when you are interviewing, instead of asking people to rate their abilities, ask them specific questions about what they have done, and then apply your own rating scale. Their opinion of their own abilities is virtually irrelevant.