Are your Dashboards Generating Facts or Furphies?

What on Earth is a Furphy?

Furphy is one of my favourite Australian slang words - according to to Wikipedia:

“A furphy is Australian slang for an erroneous or improbable story that is claimed to be factual. Furphies are supposedly 'heard' from reputable sources, sometimes secondhand or thirdhand, and widely believed until discounted.”

So what does this mean for Data Dashboards?

The best thing about data dashboards is that they help people to work with data when they don’t have any statistics training - But this is also the worst thing about dashboards. 

This leads to some dashboards that look fine to someone new to data, but are actually really misleading. There’s so much that can go wrong - manipulated axis on charts, cumulative charts, inaccurate comparisons and many more - I could write an entire article on this. 

But a key issue that’s often overlooked is how we interpret data from dashboards. Even when dashboards are carefully put together with great data, their job is to be a snapshot of what’s happening, and not tell you why it is happening. 

Context is really important - here’s a video with some more information - but what I really wanted to write about today was Correlation and Causation. 

In statistics, correlation means there’s a statistical relationship or association between two random variables or bivariate data. This differs greatly from causation, which means that one event has been caused by another event. 

So just because two variables (or lines on a chart) might look like they’re related, you can’t just say they are without proving this, because it could just be random or a coincidence. 

This is one of my favourite examples to illustrate this point, by Tyler Vigen:

So per capita cheese consumption correlates with the number of people who died by becoming tangled in their bedsheets. Does this mean that if in future people consumed less cheese, fewer people would die in their sheets?

No! 

This is just a coincidence, and it’s a fallacy (or a furphy!) to establish a cause-and-effect relationship just by looking at some charts - there’s a scientific method for this that you can’t just ignore. 

The problem is that the first question anyone asks after finding out what has happened on a dashboard is “Why?” and it’s often pretty tempting to draw your own conclusions. 

It’s crucial to make sure you’re generating facts, and not furphies, with your dashboards - especially if you’re using these dashboards as a basis for making decisions within your business. 

With this in mind, make sure you’re not over reliant on dashboards. 

I know they look nice, and I know it’s reassuring to see the numbers ticking up or down (especially when working remotely) but if you’re basing large decisions on data, make sure you’re analysing the data correctly. 

Oh, and please think of the cheese consumption graph next time you’re tempted to say correlation and causation are the same thing when you’re looking at a dashboard!

About the Author: Brooke Jamieson is Experience Lead at PlaceOS, a technology platform for creating digital experiences for buildings, workplaces, hospitals and more. Learn more about PlaceOS here and learn more about Brooke here.