3 Meaningful Ways to Visualize Margin of Error in Tableau

Now I will create another calculated field and label it Margin of Error, then enter the following calculation. 

MIN([Z-Score])*(STDEV([Discount])/SQRT(COUNT([Orders])))

This calculation will give us the margin of error if you want to display it as a single KPI. But to truly visualize it, I want the margin of error to the left and right of the Average Discount. To do this, I am going to create two more calculated fields. 

Margin of Error Left: AVG([Discount]) – MIN([Z-Score])*(STDEV([Discount])/SQRT(COUNT([Orders])))

Margin of Error Right: MIN([Z-Score])*(STDEV([Discount])/SQRT(COUNT([Orders]))) + AVG([Discount])

With our calculated fields done, we are ready to start creating our visualizations. 

How to use a rounded bar to visualize margin of error

To start, I will add AVG Discount and Measure Values to the Columns shelf, then Category and Sub-Category to the Rows shelf. When Measure Values are added to the view, you will notice that it brings in practically all of the measures into the view. 

We want to filter that list down by dragging Measure Names to the Filters shelf and then select Discount, Margin of Error Left, and Margin of Error Right then click OK. 

Now I will select the AVG(Discount) Marks card and change the Mark type to Circle.

Then select the Measure Values Marks card and change the Mark type to Line and move Measure Name to the Path property of the Marks card. 

Next, I will right-click Measure Values and choose Dual Axis.

Then I will right-click on the lower axis and choose Synchronize Axis. 

The visual we are left with is Margin of Error represented by the line and AVG Discount represented by the circles in the middle. 

I am going to clean this up a bit now by changing the Color and Size elements of the view. I will also add the AVG Discount measure to the Label property of the AVG Discount Marks card.

The Easy Way to Make Rounded Bars in Tableau

How to use reference banding

For the next visualization, I will be using a Reference Band. To start, I am going to right-click on the sheet we just made and choose Duplicate. This will carry over the formatting we just made and save us steps in the long run. 

Now I will click on the Analytics pane and drag Reference Band to the view. In the top left corner of the view, we will see a box appear with different options. We want to drag Reference Band to the Cell option on the far right, then drop it in the Measure Values Cell.

You will see some bandings appear over the rounded bar charts. To clean the view up, I am going to select the Color property from the Measure Values Marks card and change the Opacity down to zero. This will make it appear as if the rounded bars disappeared and we are left with just the bands. I will adjust the colors of the bands and call it good.

How to use box plots

Lastly, I will show you how to visualize margin of error using box plots. This approach is very similar to the last. I will duplicate the Rounded Bar sheet and then select the Analytics tab at the top of the pane. I will drag in Box Plot and drop it on the Measure Values Cell.

To clean these box plots up, I will first remove AVG Discount from the Label property of the AVG Discount Marks card. I will also decrease the size of the circle and change the color to 0% opacity. Then I will click on the Measure Values Marks card and change the line color to 0% opacity. This will leave us with just the box plots in our view. The last thing I will do is clean up the color and give it my own personal spin. 

Adding context to the analysis

We can see that some Sub-Categories have a wider margin of error than others. This is most likely due to the number of orders or sample size. For example, if we look at the Machines Sub-Category and compare that to the Paper Sub-Category; we can see that the Machine Sub-Category has a very wide margin of error compared to the much smaller margin of error for the Paper Sub-Category. If we think about this intuitively, it makes sense because we can imagine that we sell more paper than machines, so the sample size is higher for Paper. We can also imagine that machines cost more than paper products and therefore may have a wider range of discounts applied to them. 

To confirm my assumptions, let’s go back to the rounded bar chart view and add Count of Orders to the Size property of the AVG Discount Marks card and remove AVG(Discount) from the Label property. Now we can easily see that Paper has a higher order count than Machines by just looking at the size of the circle. 

I hope this tutorial helps you add more context to your dashboards and drive them to insight. 

Until Next Time,
Ethan Lang
Director, Analytics Engineering
[email protected]