|
I once chaired a personnel committee. I 'inherited' the form for evaluating the CEO, handed the 3 or 4 page thing out, and collected the results.
A third of the Board wanted to fire the CEO with cause, he'd done a horrible job; a third thought the CEO should be given a raise and a bonus, he walked on water; a third thought he was doing ok. Almost all evaluated the same man for the same 9 months that they'd seen him in action; a few had seen him in action for a couple of years. Everybody read their own definitions into the evaluation: frat boys, a self-avowed socialist, university administrators, grad students in engineering and film, a law professor and a laywer that worked in civil rights advocacy.
Two months later we fired his sorry ass for lying to the Board. We finally reached a consensus, but over an act that hadn't yet taken place when he was evaluated.
But that didn't solve the evaluation problem. I drew up a 30+ page form with an appendix stipulating exactly what each criterion meant, exactly what the basis for a "very good" versus a "good" was; my committee approved it (actually, they made it slightly longer). Then I grouped the questions, and made the Board approve the weight given to each group. When we had a new CEO, that "tool" was used. One Board member spend 5 hours on it, got half way through, and threw it in my face; that was his choice, it was thrown in the trash. When I processed all the other forms, completed, there was great unanimity. Given possible rankings between 0 and 5, and completely filling in the "placeholder" through definitions and examples, the standard deviation was something like 0.2.
|