How to Create Dumbbell Plots with R & ggplot2 | A Step-by-Step Tutorial

The best chart type to compare two values across multiple groups are dumbbell plots. They are muuuch better than the frequently used paired bar charts. In this video, I give you a step-by-step ggplot guide so that you can make better charts for comparisons.

SUPPORT THIS CHANNEL

MORE VIDEOS

MORE CONTENT

#datavisualization #ggplot2 #rstats

[toad]

My goat - your channel is a goldmine of R knowledge

[alexis]

This is amazing the production quality is so good, the cadence is perfect, thank you!!!! Your channel is a goldmine

[raould]

Thank you for this! Your videos are very helpful and I appreciate the subtle humor, besides the technical brilliance! Looking forward to many more! Good luck!

[albertocabrera]

Technical prowess in creating this dumbbell is astonishing. Before I watched this tutorial, I assumed Albert would rely on geom_dumbbell ( ). Instead, he demonstrated the use of geom_line and geom_point to produce the same result. Next, a found a most creative way of using a dummy variable whose direction changed in light of the difference in life expectancy in two years. In this regards, I was surprised that calling for a particular cell in a mutate command could actually perform the role of a function (e.g., life_exp[2]). I always assume it would be a fixed value (e.g., Albania's life expectancy in 2007 -76.423 years-). However, I found confusing splitting the mutate function for creating and ordering change life expectancy. Instead, I decided to try this variation instead : mutate(change_life_exp = diff(life_exp), .by = country,
order_dumbbells = if_else(change_life_exp < 0, -1, 1) * life_exp[2]).

[petercourtney]

Love your vids. I am so interested in creating a good way to communicate stat sig difference in means with some sort of overlapping confidence bars, but this has been surprisingly hard to derive. Any thoughts?

[Jimson-wxuq]

Hi Albert,

Can you tell me what is the best way to get nice looking conditional formatting in my tables?
I want a clear gradation in color (ie several different shades of green/red).
I want to be able to set conditions / midpoints if possible (for example, 0 always appears as white).

If it helps to clarify, my typical workflow uses Rmarkdown and KableExtra so I'm hoping to add conditonal formatting to those tables.