As I’ve said before, bar charts are good for comparing distinct values. Unlike pie charts, which only need a minimum of two values to be informative, bar charts usually have more. Usually I wouldn’t think about a bar chart unless I had at least four values to compare. It’s hard to come up with an abstract example of when one would use a bar chart, so I’m going to share the story of and instructions for the first chart I ever created that made someone say “Wow, you can do that?”
First, a little context. As anyone who works in libraries knows, books have one annoying trait – they take up space. (And they have weight, and an uncanny ability to know where your head and/or feet are when they fall, but that’s not relevant to this discussion.) Since shelves only have a fixed amount of space available, whenever a library section is first organized most libraries try to spread the books out evenly, leaving room for growth on each shelf. Unfortunately, since book size parity is for some reason not the primary concern when purchasing books, not all of that growth room is used up at the same pace. For a while the shelvers just sort of adjust by moving books up a shelf, down a shelf, or otherwise wrapping to the next shelf as needed, but eventually this becomes impossible. Either the shelf labels start to become grossly inaccurate or several shelves in a row become overstuffed, thus making it impossible to adjust on the fly.
The library I worked at had reached the point where something had to be done. They had about 150 shelves of bound journals, some of which were overstuffed, some of which had a little growth room, and a few of which had a lot. Student workers were given the task of going around and measuring the available space on each shelf, giving the library a lot of information – too much to easily interpret and use. It wasn’t a simple matter of seeing that shelf 2 was packed full, but shelf 54 has 10 inches of free space, and then moving five inches worth of journals from shelf 2 to shelf 54. That would do something like putting about one year’s worth of “American Journal of Architecture” in the middle of “JAMA (Journal of the American Medical Association)” — no one would ever be able to find it again.
This was back in ’97 or ’98 so I don’t remember the original numbers, but for this example chart I’ve just made some up.
The first thing I did, to keep things simple, was add up the total space available in each set of shelves, rather than looking at each shelf individually. After all, it’s relatively easy to physically go to a set of shelves, and say “Okay. I’ve got to move twelve inches from the bottom shelf to the top of the next set of shelves, and then adjust the remainder to have around three inches of space at the end of each shelf.” Because we can see it all at once, it’s easy to wrap our minds around. But it’s a lot harder to look at a set of shelves, and try to remember how much space there was on other sets of shelves, that you can’t see. You might remember that shelf number 21 was notably empty, but if you’re at shelf 83 and it’s completely stuffed, that doesn’t help you much.
Similarly, anyone looking at a graph with 150 elements is likely to be overwhelmed. The details get lost. It’s still a useful tool, but it could be more useful if it represented more of an overview of the situation. So to get that overview I added up the space available in each set of shelves, and used that to make a chart with 30 items, instead of one with 150.
Then, because of the way our shelves were arranged, I added some “break” lines to reflect where the line of shelves ended in one place, and moved to another place to start again. This was strictly to make it easier for people to read the chart, and translate it mentally to the physical shelves.
So, if you had a similar project, here’s how to do it.
- Measure the physical space. In this case we measured how much space was available on each shelf. Since I don’t have those numbers I just picked relatively random numbers between zero and twelve, representing up to a foot of space per shelf.
- Combine values, if appropriate. If you had, say, only 5 or 10 shelves to worry about, but they were in different rooms so you couldn’t see them all at once, there would be no need to simplify the chart. But if you make the chart, look at it and say “This doesn’t help at all! It’s still too complicated!” then you want to find some way to group the values into a more manageable number of groups. In my case I used the fact that each shelf was part of a set of five shelves, and just grouped them by set by adding up all the space available in the shelves of the set. So for set 9, the shelves had 2.3, 2.8, 4, 5.4, and 4.6 inches available. 2.3+2.8+4+5.4+4.6=44.2 inches available for set 9.
- Reflect reality. If there is a distinct break in the physical world, make a break in the graph. You can do this by copying all the items below the break and pasting them one row down, then entering the word “break” as the item label and leave the value blank.
- Make a bar chart. You can follow the instructions here.
- Add labels and formatting. You can follow the instructions here.
Now, if I were creating this chart today, I’d probably add a trendline, or at least a line showing the overall average space available, but that’s a little complicated so I’ll save it for another post. I think at the time I might have just drawn it in by hand, on the printout.