Steve Jones performs a comparison:
I ran across an interesting post from Rita Fainshtein that looked at the different types of graphs for a set of data. I thought that was interesting, so I ran my own experiment. I found for my data, a line graph was better, but let me know what you think.
My data set was simple, a few players across a few events and their number of kills. I coach volleyball and I’m always trying to present stats in a useful way. Here was the small set I picked.
Something to keep note of is that line charts generally imply time series data: you are looking at some periodic activity and analyzing changes between periods.
The stacked column (or bar) chart tells you two things: the total and the first event. It’s really hard to discern any other events from a stacked chart comparison, as Steve points out.
If you don’t have periodic data or if you’re more focused on general trends than what happened at a specific event, the Cleveland dot plot (available in custom visuals) could plot each of the events, with the Y axis representing player names and the X axis representing number of kills, and each dot being a specific game. This saves a lot of real estate and gives you an idea of how points cluster, especially if you look at a larger number of games or players.