Showing Quantile Statistics Associated with Boxplots

[#1]

Hi

I just started using JMP. When I create a boxplot through the graph builder it does not show the quantiles as shown in this example: https://www.jmp.com/content/dam/jmp/documents/en/academic/learning-library/02-box-plots-se.pdf

It only shows the boxplot. Where do I find the quantiles?

Thank you.

[#2]

Hi,

As the one-page guide shows, those box plots were created with other platforms in JMP: Analyze > Distribution, Analyze > Fit Y by X.

You can create box plots with quantiles in Graph Builder by changing the box plot type (lower left in the Graph Builder window).

I hope this helps.

Regards,

Phil

[#3]

Thank you for your answer, Phil.

I dont have problems creating  a boxplot (see my boxplot attached). I dont know how to show the quantiles (the acutal numbers, I mean).  I dont get a box with numbers. Where do I find this?

Thank you!

[#4 · Accepted solution]

If you want the values for the quantiles you will have to use a different platform in JMP. Graph Builder is great for visual analysis. For statistics look in the Analyze menu.

Again, I would refer you to the one-page guide that you mentioned in your post. This shows you how to get the quantiles using Analyze > Distribution. If you want quantiles for the variable for different levels of some other variable(s) you can use the "By" role in the launch dialogue or the local data filter.

Alternatively you could try Tables > Summary.

[#5]

I would say Distribution and then show Quantile Box Plots. The table comes from Distrubiton platform

[#6]

I think "5 Number Summary" does the trick. In graph builder after creating a box plot, right click on the graph area, choose Box Plot, then "5 Number Summary". Max, Q3, Med, Q1 and Min are then displayed on the graph.

I wish the "5 Number Summary" were actually a "6 Number Summary". I would include N, the number of data points. When presenting box plot data, this is the most common question I get, "how many data points are represented by that box?" It's a more common question than the p-value or anything else.

[#7]

The '5 number summary' was originally developed by John Tukey as part of his regular data analysis in the days before computer graphics. He later developed the 'box and whiskers plot' as the graphical representation of the same summary. So JMP is merely following the historical practice that is used by many analysts.

Your situation might be addressed by adding a plot of N for each group. That result is easily accomplished with Graph Builder:

Or perhaps include the markers for the data to get a sense of the group sizes.

The best solution depends on the particular details of your data, of course.