The graph below shows the average daily maximum temperatures (blue dots) and average daily minimum temperatures (red dots) from 2000 to 2024 in June.
From the explanation above, you should be able to get a rough idea of what this graph shows.
This graph is called a "Dumbbell Chart" because by connecting two values (e.g. the minimum and maximum temperatures) with a line, it forms a dumbbell shape.
Dumbbell charts are effective for visually representing the difference or range of fluctuation between two values , and are particularly well suited to showing changes over time or differences between groups.
In this article, we will use the "Graph Builder" in JMP toHere's how to create this dumbbell chart.
Graph Builder does not have a graph type called a "dumbbell chart." The key to creating one is to specify two numeric variables in the "interval" zone, which can be used as the lower and upper limits of the interval .
Steps to create a dumbbell chart in JMP
The JMP sample data file "Cholesterol.jmp" divides subjects into four treatment groups (A, B, control, placebo) and contains cholesterol values measured in the morning and afternoon once a month for three months.
Here, we will show the change from "April morning" to "June afternoon" on a dumbbell chart.
*The actual sample data does not have a column called "ID." This data shows an "ID" column added to distinguish subjects. This data can be downloaded from this page (Cholesterol_Dumbbell_Chart.jmp).
Operating Procedure
After launching Graph Builder, do the following:
1. Add "ID" to the [Y] zone, select "April AM" and "June PM" at the same time, and drop them into the [X] zone.
2. Select "April Morning" and "June Afternoon" simultaneously and drop them into the [Section] zone.
Now that you have created a dumbbell chart, you can customize the graph by changing the axis settings and marker settings.
You can see these steps in action in the video below.
Steps to draw a dumbbell chart (no audio)
Example of a dumbbell chart after molding
This is a dumbbell chart created by molding. From this graph, we can see that there is a tendency for cholesterol levels to decrease significantly in Groups A and B, while in the control and placebo groups there is little change or even an increase.
Note : If you want to create a graph that shows the range of values for two numeric variables, you can also draw a bar graph for the two variables and then change the Bar Style to Interval .
by Naohiro Masukawa (JMP Japan)
Naohiro Masukawa - JMP User Community
