The use of averages is a straightforward method for quantifying patterns or conditions that apply throughout an entire organization. But, at most, they may be described as a blunt tools. This is why we need to adopt a more nuanced approach and how we should go about doing it.
There is a well-known tale of a statistician who tried to ford a river based on the assumption that the water would be no deeper than six feet, as this was the depth that the statistician had determined the river to be on average. If he were here to explain it to you now, you’d understand why taking the average is such a risky proposition.
Yet, rather than retelling the events that transpired with this statistician, allow me to share some tales that are more personal and closer to home. The inclusion score of a team was scored 4+ out of a theoretical maximum of 5, according to an annual engagement survey; yet, complaints from new recruits continued flowing in, and these new hires obviously felt considerably less involved than 4+/5. It wasn’t until we compared both measurements, the average and the individual, that we understood that individuals who had been around since before the pandemic formed strong bonds with one another and were quite excellent at maintaining their connections. On the other side, this team had a much smaller number of new employees who, despite the greatest efforts, were mistakenly left out of most virtual coffee conversations and informal huddles and, as a result, felt terribly ignored. In this instance, the average score significantly misrepresented those individuals who did not feel included, while significantly overrepresenting those individuals who did feel included.
We are aware that averages provide us with very little information. And yet, it permeates our offices and the dashboards of our vehicles. For example, an average occupancy rate of 22% informs us very little about attendance patterns and much less about the floors and buildings that could have to be closed down in the future. When we try to compare various populations by using their averages, the task becomes considerably more difficult.
For our roles, averages offer an even larger hazard. It is impossible to take a look at the whole situation and deduce from a simple answer that this is how everyone is feeling. Everyone in the workplace is going through quite diverse experiences, despite the fact that the number may be perceived either as a sign that things are going well or that they are heading down the drain. We have a less firm grasp of the whole picture as a result of our attempt to reduce all of this variety to a single number. When we take these individual figures from across thousands of employees and condense them further into an average to calculate our reaction, it makes it even more difficult to obtain a clear picture.
It’s true that averaging is simple, but it may also be very deceiving. When an average is employed to represent an unknown number, the findings end up being skewed because the average disregards the influence that the inevitable fluctuations have. This causes the results to be inaccurate. If there was one movement that I could start in the realms where HR and data collide, it would be the movement to rethink how we look at averages. This is the movement that I would like to start.
What else can we look at if not the averages?
Consider adopting the path of least resistance represented by the average. We almost always fall back on averages when we are pressed for a single statistic to use as a point of reference and when inquiries are posed of us. On the other hand, the tendency is shifting.
Because of our preoccupation with statistics, over the course of the previous several years, we have moved away from direct employee involvement, such as round table conversations, ad hoc one-on-one meetings, and other similar activities. The process of complementing averages with tales is reassuring since I am seeing that there is a tendency in the other direction towards increasing connection circles and leaders holding “coffee talks” and listening circles. When there is a discrepancy between the evidence and anecdotes, one sage leader once advised their followers to always put their faith in the anecdotes. So, while the statement in question was initially designed for use in connection with product feedback, the same logic can be applied to the vast majority of situations.
But I’d still prefer some quantifiable data
In light of the abundance of information that surrounds us, I am aware that participating in a discussion with simply stories and not any numbers might be counterproductive. This is especially true considering the quantity of information that is already available. And I have never once said that information is of poor quality. My problem is not with the concept of averages in and of themselves. As a result, I will now present the best three alternatives to using bare averages.
Never state the what without explaining why: My team has a term that we repeat a bit too frequently, and it is “never state the what without explaining why.” The term “naked numbers” refers to a set of digits that, taken on their own, don’t convey a lot of information and can be misleading. If you haven’t already grasped it, my emotions against averages are the same as those I have for bare numbers. As a result, my golden rule is to never start with the what before explaining the why. When someone asks me for data, before I give them the number itself, I always explain the narrative behind the number and what it represents first, even if this upsets some people. When I do that, it compels me to evaluate it in light of other pertinent metrics and to enhance my analysis with anecdotes. When I present data in this day and age without accompanying stories, my listeners become concerned.
Put your attention on the range, and don’t worry about the mean
Here is an alternative method to consider in the event that you do not have the luxury of narrating a tale before presenting the data. Remove averages completely. Throughout the process of data collection, we already employ distributions, and we also adore graphs. One easy adjustment, such increasing the single average to represent the answer dispersion, enables the reader to distinguish between those who are on the fence and those who are against something. The fact that a training was given a score of at least four and a half out of five on a five-point scale does not tell us very much; but, the number of persons who gave each score does tell us a little bit more about how the score was ultimately determined to be four and a half. A distribution over time (for example, the demand for training) also enables us to plan our resources more effectively. If, for example, the demand for a training throughout the year was four instances on average each month but nil in the first and final quarters of the year, then allocating resources to facilitate four instances on average each month won’t do us any good. If this is the case, remove the averages from the first page and replace them with graphs and distributions instead.
There must be some universe where everything is evenly distributed and where arithmetic and statistics make perfect sense. The unfortunate reality is that world is not this one, and it is highly unlikely that it ever will be. When it comes to formulating plans for the reality in which we now exist, a statistic as straightforward and obscure as averages is a poor choice to base one’s decisions on. Just ask those who have been given the exclusive responsibility of reporting the average salary disparity between men and women. We live in a world that is overflowing with data, and it is time for us to become more knowledgeable about what we can do with it and how to put it to use.
Hence, the next time you hear someone talking about averages, tell them the story of the statistician and that you’d like to prevent the next one from drowning in their own statistics. Let’s reevaluate that arithmetic mean.
