Or better yet, when would YOU use one?
I think to all of the data displays we were taught in school and with box-and-whiskers graphs I think I’ve only seen two used outside of classes. And I have seen data that could have used a BaW graph but didn’t. And stem-and-leaf plots. Has anyone ever seen on for real?
Any other charts that you remember from school that at the time was presented as the most important thing ever and turned out to be worthless?
I use “spaghetti charts” like this one all the time:
In my case, it’s for work. But it’s usually to get a feel for the underpinnings of what various computer programs are doing for us in the typical day-to-day usage.
Examples of box and whiskers plot - Google Image Search.
I have not used or seen strictly those identify sample quartiles. But a close variant is commonly used to plot securities prices with the daily open and close as the box and the intra-day high and low as the whiskers.
Examples of stem and leaf plot - Google Search.
Those are not something I’ve much used or seen. The underlying data structure they represent (a tree) is used all over the place in computer programming, but you rarely need to display the data in an S&L-style ragged table.
Sometimes when I’m giving a science talk, I dig back into the footnotes to find the original papers about the subject I’m discussing, and those sometimes have box & whiskers, I think. Can’t find an example at the moment.
I see and use box plots quite frequently. I do a mix of engineering, science, and data science types of work. Using a box plot to view a single distribution is not particularly useful versus a histogram or something similar. However, to compare groups of distributions, boxplots are pretty valuable.
Their heyday may have passed. Wikipedia says
I would add that modern computers have made it much easier to work with relatively large data sets, for which stem-and-leaf plots aren’t well suited.
I’ve never heard of either.
I have an engineering degree.
Well said.
S&L plots are a visually pleasing way to enumerate the contents of a list of buckets. When each bucket holds ~10K items, and there are 100+ buckets, the “pleasing” part gets kinda threadbare.
Much mo betta nowadays to do a 3D countoured surface colored plot.
We use box and whiskers charts to show performance on certain financial metrics across stores.
e.g. food waste may be somewhat random across stores. If the median moves by more than the inter-quartile range either up or down in a few months it gives you a visual indicator that something real has changed.
But it’s definitely considered old school.
I see box-and-whiskers plots all the time in scientific publications, although I don’t remember ever learning about them in school. I’m not familiar with stem-and-leaf ones.
I use box and whisker plots. Mostly just showing the range of a large amount of environmental data we have collected. For instance, my team has collected water quality samples 24 times in the last year at multiple locations. We use the B&W plots to show the difference in range from say upstream locations to downstream locations.
I’ve used box-and-whiskers plots. They’re good when sometimes the mean values are interesting, sometimes the dispersions are interesting, and sometimes the ranges are interesting.
An old friend described her graduate work in stellar statistical distributions. Punched data cards were new then. She was setting a sorting machine to sort by the least significant digit, running her data cards through, assembling the results into one stack, then setting the machine to sort by the next to least significant, and so on. The output stacks of sorted cards eventually looked like a histogram, but the steps near the end showed some interesting patterns too. Stem and leaf plots remind me of her experience.
Yeah, I’ve used structures that resemble stem and leaf in databases, but never shown to the end user in that form.
How timely. I’m tutoring a stats class at a local prison, and this week they are working with box and whiskers charts.
I don’t think I’ve ever used a stem and leaf chart, and had to click the link to see what it is. But i see box and whiskers charts all the time, mostly in medical papers. All those tests of how effective the various vaccines were? They had box and whiskers charts showing the antibody levels, t cells, etc., of the research subjects.
I used box and whiskers charts when i surveyed my college class about all sorts of stuff. I used them to represent the ages of kids, split by how many kids a classmate had. I mostly use other charts, though.
I don’t think i need to tell my students that those charts are useless, but maybe you’ll convince me. 
I have done a ton of graphs over the years and this is the first time I’ve seen these terms. And they are ridiculous.
The box thing is better done with a plot with error bars.
The other is called “a table”. I know, really deep.
It depends how much data you have. If you can see all the points, yes, a plot is better. But with enough data points to make that messy, the box chart can be good.
It also depends on your audience, of course.
Don’t the boxes cover up more stuff than dots?
One box can show an arbitrary number of underlying data points, though.
Actually, it can hide an arbitrary number of data points. And that’s not good.
I find box and whiskers charts to be a helpful way to summarize data that is otherwise unwieldy. If you have few enough data points to actually see them all, sure, that’s a reasonable way to display the data, too. Sometimes it’s better, sure. But often, if you try to print a dot for every data point, you just get an unintelligible blur.