This following post by Josh Bersin (one of Reid Hoffman’s co-authors of “The Alliance” which I covered in the previous post) made the rounds a few months back and is still top-of-mind for me:
The gist: most common HR practices (performance reviews, promotions, etc.) pre-suppose a normal-distribution of performance across the employee base. However, careful studies have shown that real performance looks more like a power-distribution than a normal distribution. Many of the common HR practices “break” under those assumptions.
Bersin provides references to the academic papers that drew this conclusion, yet none of them hypothesizes why performance looks the way it does. Here’s my thesis: in the general/broader population, performance/mastery of a given skill does distribute normally. However, in almost any professional setting, we apply some sort of screening process – we intentionally try to hire the people who are “above average” (if not the top 5-10%) who are most qualified for the position. Since we don’t just pick people at random, there’s no reason to expect a normal performance distribution post-screening. We would expect to see something that resembles the right-most portion of a normal distribution – which also looks a lot like a power distribution.
These articles make a compelling case for why the current HR practices don’t make a lot of sense under these assumptions. But they fail to propose a prescriptive way to change them in the right direction.
What do you think? If we assume hat performance (at least in professional settings) follow a power-distribution, how should we change some of the key HR practices?
Love the hypothesis here. It’d be interesting to look at data from equally sized companies that have differing levels of selectiveness with their employees. A Walmart vs. a Google for example likely have very different distributions. Hard part would be a system for how to objectively compare across very different companies in different industries.
LikeLike
[…] Wayne Lin on The Bell Curve Is A Myth […]
LikeLike
[…] Avoid rating on a “bell curve” — since human performance does not seem to follow a bell curve. […]
LikeLike