A couple of weeks ago I came across a really interesting post on LinkedIn and I did what I always do in these situations. I read the post a couple of times, and then went looking for the original article or source for the LinkedIn post. I found it easily enough.
The post in question was about this much-discussed idea of ‘the normal curve of performance‘. It has been a constant in conversations about performance, particularly at performance review time, for about 20 years or so – at least that’s as far back as I can remember it showing up.
The idea is simple enough. Human performance ‘fits’ the so-called normal distribution curve. End of explanation. Nobody ever questions this ‘fact’ and so it has been used to make all kinds of decisions about the distribution of incentives, bonuses, promotions and to decide who ‘actually’ is a top performer. Year on year managers are forced to fit their assessment of their people into this normal distribution curve. I mean if it’s a fact and if everyone seems to agree on the fact then what’s the problem.
I cast my mind back to the last time I was forced to fit my people into this distribution curve. It was 2008 and before that 2007, 2006, 2005 and 2004. Before 2004 there wasn’t such pressure to do this fitting of people but I’m not sure why. In any event I sat down and looked at my team and assessed who was giving it everything and who was taking it easy. I didn’t find that my ratings matched any kind of normal distribution curve but I was told to make them fit. This felt very wrong in 2004, 2005, 2006 and every year after. It still feels wrong.
So, on to the subject of this article. Where does the idea that human performance mirrors the normal distribution curve come from? Loads of scientific research? Peer-reviewed and rigorous research? Tested and repeated time and again and confirming the same conclusions? Nope. None of that. It comes from a statement in an article in 1947. Yes, a statement that simply said ‘ratings for a large and representative group of assistant managers should be distributed in accordance with the percentages predicted for a normal distribution’. Hmm. Anything else to back up this assertion or instruction? No.
In the absence of data to contradict this idea then perhaps we’re best to stick with it, right? I mean if it’s the best working theory then we are best to stick with it absent any contradictory evidence.. right? Wrong. You see that’s the thing. There is contradictory evidence and it has been around for a few years now. And it’s very compelling, in fact it’s highly compelling. It suggests that this notion of human performance fitting the normal distribution curve is utterly ridiculous, and it proves why. Better still, it suggests that there is in fact a better profile for the likely shape of human performance. The shape of this profile is closer to what’s called a Power Law or paretian distribution. Without jumping completely onto this idea and saying that it is the absolute truth of human performance, I spent the time reading all the research I could get my hands on and it’s amazing.
I pondered whether to share this new insight with my colleagues in the past week or two. Honestly I thought they wouldn’t be able to handle what this is saying. It shatters much of what is taken as read in HR and organisational behaviour theory – and frankly that’s a good thing because much of it is so wrong. I laboured over discussing it and against my better judgment I decided to bring it up last week. I shouldn’t have bothered. I explained clearly and simply what the latest data are saying. Blank expressions and attempts to move the conversation on were what I encountered.
This won’t go away. Our love affair with the normal distribution curve as a way of fitting people’s performance into neat little cost-managed buckets is doomed. Maybe not this performance review cycle, maybe not even the next, but it is doomed. Maybe then we’ll start assessing people on the basis of what they actually achieved, how they actually performed, and we’ll finally feel a whole lot more honest about the process we’re using.
To close, I want to apologise to my team for those years where I dished out ratings you and I both knew didn’t mirror what you delivered or what I truly believed about your contribution. I know better now and I know why now.
Justin's 'Just One Thing..' 
