How do you explain the importance of effect size to non-statistical audiences?

Effect size is like a magnifying glass for your data, showing how much of an impact one variable has on another. It helps you understand the strength of a relationship, not just whether it exists.

In my experience, effect size comes into play when trying to gauge the apparent effectiveness of a decision over time or estimating future outcomes. For the past, I tend to use dual axis stacked column charts with lines or reference marks. During a specific exploratory analysis project, I had to demonstrate that decision [x] implemented on [y] date led to a decrease in [z] over time. For projections, I tend toward using line or scatter charts - using past averages at certain intervals for [a] variable and [b] population as a baseline, you can show how those might be projected for changes in either [a] or [b]. Correlation isn't causality, but in these instances, stakeholders usually want you to consider all factors are constant.

Think of effect size as the difference between knowing there's a change and understanding how big that change is. For example, it’s the difference between knowing a new marketing strategy works and knowing it increases sales by 20%.

I would caution on using real-world terms too broadly. In my experience, audiences usually want the visualization (or supplementary tex) to reference the colloquial terminology the business uses. For example, as opposed to language that uses only '[x] population grew [y]% over [z] months', it may be better to say that '[x] population grew/contracted by an anticipated/unanticipated amount for normal, seasonal expectations'. You may need to get feedback from and constantly collaborate with stakeholders and subject matter experts, but this may achieve better, more impactful results.

If your data allows you to find variable coefficients, and you can express that as a line on x-y axis alongside the other variables, then this is a great way to demonstrate to an audience if decisions or changes were impactful or not. Alternatively, you can take the percentages increases/decreases of populations under consideration while displaying the absolute numbers as either bar or column charts to demonstrate effect (if you're audience is not well-versed in statistics, then this may be the better route to take).

Effect size measures the real impact of an intervention. It’s crucial for making informed decisions because it tells you whether a change is significant enough to matter in practice.

In business intelligence, understanding effect size helps in prioritizing actions. If you know the magnitude of the impact, you can focus on strategies that yield the greatest benefit, ensuring resources are used efficiently.

An image of a lever with the fulcrum near the left end can visually show how a small movement at that side results in a very large movement on the opposite, right end of the lever. One can also populate the visual image with actual numbers (e.g., an increase in x units or percentage applied here results in an effect size increase of z units or percentage). You'll need the statistics to substantiate the actual numeric quantification of the effect size, but the visualization of a lever with the fulcrum near one endpoint helps to communicate the concept of effect size.

Always consider the context and the practical implications of the effect size. A small effect size in a critical area can be more important than a large effect size in a less relevant area.

I remember my Professor’s explanation: What’s your favourite Ice Cream? You have only two options: Chocolate and Strawberry. — The class scores the favourite on a scale of 1-10 Analysis: — Small Effect Size: If chocolate scores 9 out of 10 and strawberry scores 9.2 out of 10, the difference is small. Interpretation: Manufacture both! Sell both! — Big Effect Size: If chocolate scores 8 out of 10 and strawberry scores 2 out of 10, the difference is big. Obviously chocolate is much more popular! Interpretation: Manufacture more chocolate and possibly stop selling strawberry. In a nutshell: Effect size a simple but powerful metrics as it not just points to the difference, but how big and important that difference is.